From 6551c7f46dba18426bd3bca8b7430469d9ee44e0 Mon Sep 17 00:00:00 2001
From: Pieter Verschaffelt <pieter.verschaffelt@ugent.be>
Date: Wed, 5 Oct 2022 15:59:38 +0200
Subject: [PATCH 1/6] Update and change the way that the taxa filter works

---
 .../custom-database/CreateCustomDatabase.vue  |  75 +++++++++-
 src/components/proteome/ProteomeBrowser.vue   |  17 +++
 src/components/taxon/TaxaBrowser.vue          | 138 +++++++++++-------
 .../metadata/MetadataCommunicator.ts          |   6 +-
 src/logic/communication/proteomes/Proteome.ts |  14 ++
 .../proteomes/ProteomeCommunicator.ts         |  38 +++++
 6 files changed, 224 insertions(+), 64 deletions(-)
 create mode 100644 src/components/proteome/ProteomeBrowser.vue
 create mode 100644 src/logic/communication/proteomes/Proteome.ts
 create mode 100644 src/logic/communication/proteomes/ProteomeCommunicator.ts

diff --git a/src/components/custom-database/CreateCustomDatabase.vue b/src/components/custom-database/CreateCustomDatabase.vue
index 90c1c2e..b832495 100644
--- a/src/components/custom-database/CreateCustomDatabase.vue
+++ b/src/components/custom-database/CreateCustomDatabase.vue
@@ -33,7 +33,7 @@
                     </v-stepper-step>
 
                     <v-stepper-content step="1">
-                        <v-form ref="databaseForm">
+                        <v-form ref="databaseDetailsForm">
                             <v-container fluid>
                                 <v-row>
                                     <v-col cols="12">
@@ -51,6 +51,52 @@
                                             v-model="databaseName">
                                         </v-text-field>
                                     </v-col>
+                                </v-row>
+                                <v-row>
+                                    <v-col cols="12" class="pb-0">
+                                        <span class="font-weight-bold">
+                                            Please choose how you want to construct your custom protein reference database.
+                                        </span>
+                                    </v-col>
+                                    <v-col cols="6" class="d-flex">
+                                        <span>
+                                            Manually select which UniProt sources (e.g. TrEMBL and SwissProt) should be
+                                            used for the database construction and which proteins should be included based
+                                            on a given set of taxa.
+                                        </span>
+                                    </v-col>
+                                    <v-col cols="6" class="d-flex">
+                                        <span>
+                                            Provide a list of UniProt reference proteomes that should be used as the basis
+                                            for a custom protein reference database. All available UniProt sources (both
+                                            TrEMBL and SwissProt) will be consulted in this case.
+                                        </span>
+                                    </v-col>
+                                </v-row>
+                                <v-row>
+                                    <v-col cols="6" class="d-flex flex-column">
+                                        <v-btn @click="manuallyFilterDatabase">
+                                            Manually filter database
+                                        </v-btn>
+                                    </v-col>
+                                    <v-col cols="6" class="d-flex flex-column">
+                                        <v-btn @click="filterByProteomes">
+                                            Construct from reference proteomes
+                                        </v-btn>
+                                    </v-col>
+                                </v-row>
+                            </v-container>
+                        </v-form>
+                    </v-stepper-content>
+
+                    <v-stepper-step step="2" :complete="currentStep > 2">
+                        Database details
+                    </v-stepper-step>
+
+                    <v-stepper-content step="2">
+                        <v-form ref="databaseSourcesForm">
+                            <v-container fluid>
+                                <v-row>
                                     <v-col cols="6">
                                         <v-select
                                             dense
@@ -85,19 +131,19 @@
                                 </v-row>
                                 <v-row>
                                     <v-col cols="12">
-                                        <v-btn color="primary" @click="validateAndContinue">Continue</v-btn>
+                                        <v-btn color="primary" @click="selectDatabaseSources">Continue</v-btn>
                                     </v-col>
                                 </v-row>
                             </v-container>
                         </v-form>
                     </v-stepper-content>
 
-                    <v-stepper-step step="2" :complete="currentStep > 2">
+                    <v-stepper-step step="3" :complete="currentStep > 3">
                         Filter
                         <small>Select which organisms will be present in the output database</small>
                     </v-stepper-step>
 
-                    <v-stepper-content step="2">
+                    <v-stepper-content step="3">
                         <v-container>
                             <v-row>
                                 <v-col cols="12">
@@ -136,6 +182,7 @@ import { Prop, Watch } from "vue-property-decorator";
 import CachedNcbiResponseCommunicator from "@/logic/communication/taxonomic/ncbi/CachedNcbiResponseCommunicator";
 import ConfigurationManager from "@/logic/configuration/ConfigurationManager";
 import CustomDatabase from "@/logic/custom_database/CustomDatabase";
+import ProteomeCommunicator from "@/logic/communication/proteomes/ProteomeCommunicator";
 
 @Component({
     components: { TaxaBrowser }
@@ -226,6 +273,10 @@ export default class CreateCustomDatabase extends Vue {
                 .replace("_", ".")
 
             this.selectedVersion = lastVersion;
+
+            const results = await ProteomeCommunicator.getProteomeById("UP000464341");
+            console.log(results);
+
             this.loading = false;
         } catch (e) {
             console.error(e);
@@ -233,13 +284,25 @@ export default class CreateCustomDatabase extends Vue {
         }
     }
 
-    private validateAndContinue(): void {
+    private manuallyFilterDatabase(): void {
         // @ts-ignore
-        if (this.$refs.databaseForm.validate()) {
+        if (this.$refs.databaseDetailsForm.validate()) {
+            // this.selectProteomes = false;
             this.currentStep = 2;
         }
     }
 
+    private filterByProteomes(): void {
+
+    }
+
+    private selectDatabaseSources(): void {
+        // @ts-ignore
+        if (this.$refs.databaseSourcesForm.validate()) {
+            this.currentStep = 3;
+        }
+    }
+
     private async buildDatabase(): Promise<void> {
         // No filtering should be applied in this case (which means we pass only the root to the construction step of
         // the database).
diff --git a/src/components/proteome/ProteomeBrowser.vue b/src/components/proteome/ProteomeBrowser.vue
new file mode 100644
index 0000000..524f73e
--- /dev/null
+++ b/src/components/proteome/ProteomeBrowser.vue
@@ -0,0 +1,17 @@
+<template>
+
+</template>
+
+<script lang="ts">
+import Vue from "vue";
+import Component from "vue-class-component";
+
+@Component
+export default class ProteomeBrowser extends Vue {
+
+}
+</script>
+
+<style scoped>
+
+</style>
diff --git a/src/components/taxon/TaxaBrowser.vue b/src/components/taxon/TaxaBrowser.vue
index f457b50..c59d06d 100644
--- a/src/components/taxon/TaxaBrowser.vue
+++ b/src/components/taxon/TaxaBrowser.vue
@@ -25,66 +25,79 @@
                         {{ failedImports.join(", ") }}.
                     </v-alert>
                 </div>
-                <div v-if="selectedItems.length === 0" style="text-align: center">
-                    <div>No taxa selected yet. No filtering will be applied.</div>
-                    <div class="text-caption">
-                        Use the search bar below to find taxa that can be used for filtering.
-                    </div>
-                </div>
-                <v-chip-group column v-else>
-                    <v-chip
-                        v-for="taxon in selectedItems"
-                        close
-                        :key="taxon.id"
-                        @click:close="selectItem(taxon)"
-                        :color="getRankColor(taxon.rank)"
-                        dark>
-                        {{ taxon.name }}
-                    </v-chip>
-                </v-chip-group>
-                <div class="d-flex flex-row justify-center my-2">
-                    <v-tooltip bottom open-delay="500">
-                        <template v-slot:activator="{ on, attrs }">
-                            <v-btn
-                                color="primary"
-                                outlined
-                                v-on="on"
-                                @click="importTaxaFromFile"
-                                :loading="importLoading"
-                                class="mr-2">
-                                Import taxa from file
-                            </v-btn>
-                        </template>
-                        <span>Import a selection of taxa for filtering from a file.</span>
-                    </v-tooltip>
-                    <v-tooltip bottom open-delay="500">
-                        <template v-slot:activator="{ on, attrs }">
-                            <v-btn outlined v-on="on" @click="clearSelection" color="red" class="mr-2">
-                                Clear selection
-                            </v-btn>
-                        </template>
-                        <span>Clear selection</span>
-                    </v-tooltip>
-                </div>
-                <div>
-                    <span v-if="uniprotRecordsLoading">Computing database size...</span>
-                    <span v-else>Resulting database will contain {{ formattedUniprotRecords }} UniProtKB records.</span>
-                </div>
+
+                <v-container fluid>
+                    <v-row class="d-flex align-center">
+                        <div class="flex-grow-1">
+                            <div v-if="selectedItems.length === 0" style="text-align: center">
+                                <div>No taxa selected yet. No filtering will be applied.</div>
+                                <div class="text-caption">
+                                    Use the table and search bar below to find taxa that can be used for filtering.
+                                </div>
+                            </div>
+                            <v-chip-group column v-else>
+                                <v-chip
+                                    v-for="taxon in selectedItems"
+                                    close
+                                    :key="taxon.id"
+                                    @click:close="selectItem(taxon)"
+                                    :color="getRankColor(taxon.rank)"
+                                    dark>
+                                    {{ taxon.name }}
+                                </v-chip>
+                            </v-chip-group>
+                        </div>
+                        <div class="d-flex flex-column justify-center">
+                            <v-tooltip bottom open-delay="500">
+                                <template v-slot:activator="{ on, attrs }">
+                                    <v-btn
+                                        color="primary"
+                                        outlined
+                                        v-on="on"
+                                        @click="importTaxaFromFile"
+                                        :loading="importLoading"
+                                        class="mb-1"
+                                        small>
+                                        Import taxa from file
+                                    </v-btn>
+                                </template>
+                                <span>Import a selection of taxa for filtering from a file.</span>
+                            </v-tooltip>
+                            <v-tooltip bottom open-delay="500">
+                                <template v-slot:activator="{ on, attrs }">
+                                    <v-btn outlined v-on="on" @click="clearSelection" color="red" small>
+                                        Clear selection
+                                    </v-btn>
+                                </template>
+                                <span>Clear all selected taxa.</span>
+                            </v-tooltip>
+                        </div>
+                    </v-row>
+                    <v-row>
+                        <div>
+                            <span v-if="uniprotRecordsLoading">Computing database size...</span>
+                            <span v-else>
+                                Resulting database will contain {{ formattedUniprotRecords }} UniProtKB records.
+                            </span>
+                        </div>
+                    </v-row>
+                </v-container>
+
             </div>
         </div>
 
         <div>
-            <div>
-                <!-- Must use append-icon instead of clearable here, otherwise model is set to null -->
-                <v-text-field
-                    prepend-icon="mdi-magnify"
-                    label="Search"
-                    v-model="search"
-                    append-icon="mdi-close"
-                    @keydown.enter="filterByName()"
-                    @click:append="clearFilter()">
-                </v-text-field>
-            </div>
+<!--            <div>-->
+<!--                &lt;!&ndash; Must use append-icon instead of clearable here, otherwise model is set to null &ndash;&gt;-->
+<!--                <v-text-field-->
+<!--                    prepend-icon="mdi-magnify"-->
+<!--                    label="Search"-->
+<!--                    v-model="search"-->
+<!--                    append-icon="mdi-close"-->
+<!--                    @keydown.enter="filterByName()"-->
+<!--                    @click:append="clearFilter()">-->
+<!--                </v-text-field>-->
+<!--            </div>-->
             <v-data-table
                 :headers="headers"
                 :items="taxa"
@@ -93,6 +106,19 @@
                 :loading="loading"
                 :options.sync="options"
                 dense>
+                <template v-slot:footer.prepend>
+                    <v-text-field
+                        prepend-icon="mdi-magnify"
+                        label="Search"
+                        v-model="search"
+                        append-icon="mdi-close"
+                        dense
+                        hide-details
+                        class="mr-6"
+                        @keydown.enter="filterByName()"
+                        @click:append="clearFilter()">
+                    </v-text-field>
+                </template>
                 <template v-slot:item.action="{ item }">
                     <v-btn
                         color="primary"
diff --git a/src/logic/communication/metadata/MetadataCommunicator.ts b/src/logic/communication/metadata/MetadataCommunicator.ts
index e232861..90684d1 100644
--- a/src/logic/communication/metadata/MetadataCommunicator.ts
+++ b/src/logic/communication/metadata/MetadataCommunicator.ts
@@ -33,11 +33,13 @@ export default class MetadataCommunicator {
         swissprotSelected: boolean,
         tremblSelected: boolean
     ): Promise<number> {
+        let idQuery: string;
         if (taxa.length === 0) {
-            return 0;
+            idQuery = "*";
+        } else {
+            idQuery = taxa.map(taxon => `taxonomy_id:${taxon}`).join("+OR+");
         }
 
-        const idQuery = taxa.map(taxon => `taxonomy_id:${taxon}`).join("+OR+");
         const result = await NetworkUtils.getJSON(
             `${MetadataCommunicator.UNIPROT_API_URL}?facets=reviewed&query=${idQuery}&size=0`
         );
diff --git a/src/logic/communication/proteomes/Proteome.ts b/src/logic/communication/proteomes/Proteome.ts
new file mode 100644
index 0000000..2649e99
--- /dev/null
+++ b/src/logic/communication/proteomes/Proteome.ts
@@ -0,0 +1,14 @@
+import { NcbiId } from "unipept-web-components";
+
+export default class Proteome {
+    constructor(
+        // String that uniquely identifies this reference proteome.
+        public readonly id: string,
+        // Name of the organism to which this proteome is associated.
+        public readonly organismName: string,
+        // NCBI ID of the organism to which this proteome is associated.
+        public readonly organismId: NcbiId,
+        // The amount of proteins that are present in this reference proteome.
+        public readonly proteinCount: number
+    ) {}
+}
diff --git a/src/logic/communication/proteomes/ProteomeCommunicator.ts b/src/logic/communication/proteomes/ProteomeCommunicator.ts
new file mode 100644
index 0000000..d119b54
--- /dev/null
+++ b/src/logic/communication/proteomes/ProteomeCommunicator.ts
@@ -0,0 +1,38 @@
+import Proteome from "@/logic/communication/proteomes/Proteome";
+import { NetworkUtils } from "unipept-web-components";
+
+export default class ProteomeCommunicator {
+    public static async listReferenceProteomes(): Promise<Proteome[]> {
+        return null;
+        //
+        // const result = await NetworkUtils.getJSON(
+        //     "https://rest.uniprot.org/proteomes/stream?fields=upid%2Corganism%2Corganism_id%2Cprotein_count%2Cbusco%2Ccpd&format=tsv&query=%28%2A%29"
+        // );
+        //
+        // console.log(result);
+        // return null;
+    }
+
+    public static async getProteomeById(id: string): Promise<Proteome> {
+        const result = await NetworkUtils.get(
+            `https://rest.uniprot.org/proteomes/stream?compressed=false&fields=upid%2Corganism%2Corganism_id%2Cprotein_count&format=tsv&query=${id}`
+        );
+
+        const resultLines = result.trimEnd().split("\n");
+
+        if (resultLines.length < 2) {
+            // Only the header is present, we didn't find the requested proteome and return null.
+            return null;
+        } else if (resultLines.length === 2) {
+            // We found the requested proteome and will return it.
+
+            // Input data is TSV, so we split the fields by a tab-character.
+            const data = resultLines[1].split("\t");
+            // First field is ID, second is organism name, third is organism ID, fourth is protein count
+            return new Proteome(data[0], data[1], parseInt(data[2]), parseInt(data[3]));
+        } else {
+            // Multiple reference proteomes with this ID were found and an error will be thrown in this case.
+            throw new Error(`Invalid reference proteome ID ${id} was provided.`);
+        }
+    }
+}

From a37444382071ce9d6cf30db04cd37e8cd5a4cb08 Mon Sep 17 00:00:00 2001
From: Pieter Verschaffelt <pieter.verschaffelt@ugent.be>
Date: Thu, 6 Oct 2022 16:11:24 +0200
Subject: [PATCH 2/6] Reference Proteomes can be processed now, not yet revived

---
 .../custom-database/CreateCustomDatabase.vue  | 341 +++++++++++++-----
 src/components/taxon/TaxaBrowser.vue          |  22 +-
 src/logic/AsyncHelper.ts                      |  41 +++
 .../docker/DockerCommunicator.ts              |   4 +-
 .../communication/uniprot/UniProtConstants.ts |   6 +
 .../analysis/AnalysisSourceManager.ts         |   2 +
 .../docker/CustomDatabaseManager.ts           |   6 +
 src/state/CustomDatabaseStoreFactory.ts       |  25 +-
 8 files changed, 325 insertions(+), 122 deletions(-)
 create mode 100644 src/logic/AsyncHelper.ts
 create mode 100644 src/logic/communication/uniprot/UniProtConstants.ts

diff --git a/src/components/custom-database/CreateCustomDatabase.vue b/src/components/custom-database/CreateCustomDatabase.vue
index b832495..d3d0806 100644
--- a/src/components/custom-database/CreateCustomDatabase.vue
+++ b/src/components/custom-database/CreateCustomDatabase.vue
@@ -27,13 +27,13 @@
             </v-card-text>
             <v-card-text v-else class="mt-2">
                 <v-stepper v-model="currentStep" vertical flat>
-                    <v-stepper-step step="1" :complete="currentStep > 1">
-                        Database details
+                    <v-stepper-step step="1" :complete="currentStep > 1" :editable="currentStep > 1">
+                        Database name
                         <small>Provide basic construction details</small>
                     </v-stepper-step>
 
                     <v-stepper-content step="1">
-                        <v-form ref="databaseDetailsForm">
+                        <v-form ref="databaseNameForm">
                             <v-container fluid>
                                 <v-row>
                                     <v-col cols="12">
@@ -53,117 +53,210 @@
                                     </v-col>
                                 </v-row>
                                 <v-row>
-                                    <v-col cols="12" class="pb-0">
-                                        <span class="font-weight-bold">
-                                            Please choose how you want to construct your custom protein reference database.
-                                        </span>
+                                    <v-col cols="12">
+                                        <v-btn color="primary" @click="validateDatabaseNameAndContinue()">
+                                            Continue
+                                        </v-btn>
                                     </v-col>
-                                    <v-col cols="6" class="d-flex">
+                                </v-row>
+                            </v-container>
+                        </v-form>
+                    </v-stepper-content>
+
+                    <v-stepper-step step="2" :complete="currentStep > 2" :editable="currentStep > 2">
+                        Construction method
+                        <small>Please select how you want to construct the database</small>
+                    </v-stepper-step>
+
+                    <v-stepper-content step="2">
+                        <v-container fluid>
+                            <v-row>
+                                <v-col cols="6" class="d-flex">
                                         <span>
                                             Manually select which UniProt sources (e.g. TrEMBL and SwissProt) should be
-                                            used for the database construction and which proteins should be included based
-                                            on a given set of taxa.
+                                            used for the database construction and which proteins should be included
+                                            based on a given set of taxa.
                                         </span>
-                                    </v-col>
-                                    <v-col cols="6" class="d-flex">
+                                </v-col>
+                                <v-col cols="6" class="d-flex">
                                         <span>
-                                            Provide a list of UniProt reference proteomes that should be used as the basis
-                                            for a custom protein reference database. All available UniProt sources (both
-                                            TrEMBL and SwissProt) will be consulted in this case.
+                                            Provide a list of UniProt reference proteomes that should be used as the
+                                            basis for a custom protein reference database. All available UniProt sources
+                                            (both TrEMBL and SwissProt) will be consulted in this case.
                                         </span>
+                                </v-col>
+                            </v-row>
+                            <v-row>
+                                <v-col cols="6" class="d-flex flex-column">
+                                    <v-btn @click="manuallyFilterDatabase">
+                                        Manually filter database
+                                    </v-btn>
+                                </v-col>
+                                <v-col cols="6" class="d-flex flex-column">
+                                    <v-btn @click="filterByProteomes">
+                                        Construct from reference proteomes
+                                    </v-btn>
+                                </v-col>
+                            </v-row>
+                        </v-container>
+                    </v-stepper-content>
+
+                    <div v-if="useProteomeFilter">
+                        <v-stepper-step step="3" :complete="currentStep > 3" :editable="currentStep > 3">
+                            Select reference proteomes
+                            <small>Decide on a set of reference proteomes that should be present in the database</small>
+                        </v-stepper-step>
+
+                        <v-stepper-content step="3">
+                            <v-container fluid>
+                                <v-row>
+                                    <v-col cols="12" class="d-flex">
+                                        <v-text-field
+                                            dense
+                                            label="Reference proteome identifier"
+                                            hint="Provide a unique UniProtKB reference proteome identifier (e.g. UP000005640)"
+                                            persistent-hint
+                                            :error="referenceProteomeError !== ''"
+                                            :error-messages="referenceProteomeError"
+                                            v-model="referenceProteome">
+                                        </v-text-field>
+                                        <v-btn color="primary ml-2" @click="addReferenceProteome()">Add</v-btn>
                                     </v-col>
                                 </v-row>
                                 <v-row>
-                                    <v-col cols="6" class="d-flex flex-column">
-                                        <v-btn @click="manuallyFilterDatabase">
-                                            Manually filter database
-                                        </v-btn>
+                                    <v-col cols="12">
+                                        <v-simple-table v-if="referenceProteomes.length > 0">
+                                            <template v-slot:default>
+                                                <thead>
+                                                <tr>
+                                                    <th class="text-left">Proteome ID</th>
+                                                    <th class="text-left">Organism name</th>
+                                                    <th class="text-left">Protein count</th>
+                                                    <th class="text-center">Actions</th>
+                                                </tr>
+                                                </thead>
+                                                <tbody>
+                                                <tr v-for="(proteome, index) in referenceProteomes" :key="index">
+                                                    <td>{{ proteome.id }}</td>
+                                                    <td>{{ proteome.organismName }}</td>
+                                                    <td>{{ proteome.proteinCount }}</td>
+                                                    <td class="text-center">
+                                                        <v-tooltip bottom open-delay="500">
+                                                            <template v-slot:activator="{ on }">
+                                                                <v-btn
+                                                                    icon
+                                                                    @click="removeReferenceProteome(index)"
+                                                                    v-on="on"
+                                                                    color="red">
+                                                                    <v-icon>mdi-delete</v-icon>
+                                                                </v-btn>
+                                                            </template>
+                                                            <span>Remove reference proteome from selection</span>
+                                                        </v-tooltip>
+                                                    </td>
+                                                </tr>
+                                                </tbody>
+                                            </template>
+                                        </v-simple-table>
+                                        <div v-else>
+                                            No reference proteomes have been added yet. Find one in the search bar above
+                                            and click the "Add" button.
+                                        </div>
                                     </v-col>
-                                    <v-col cols="6" class="d-flex flex-column">
-                                        <v-btn @click="filterByProteomes">
-                                            Construct from reference proteomes
+                                </v-row>
+                                <v-row>
+                                    <v-col cols="12">
+                                        <v-btn class="mr-1" @click="currentStep--">Go back</v-btn>
+                                        <v-btn color="primary" @click="buildDatabaseFromReferenceProteomes()">
+                                            Build database
                                         </v-btn>
                                     </v-col>
                                 </v-row>
                             </v-container>
-                        </v-form>
-                    </v-stepper-content>
-
-                    <v-stepper-step step="2" :complete="currentStep > 2">
-                        Database details
-                    </v-stepper-step>
-
-                    <v-stepper-content step="2">
-                        <v-form ref="databaseSourcesForm">
+                        </v-stepper-content>
+                    </div>
+
+                    <div v-else>
+                        <v-stepper-step step="3" :complete="currentStep > 3" :editable="currentStep > 3">
+                            Database sources
+                            <small>Select which UniProtKB components should be processed for the database</small>
+                        </v-stepper-step>
+
+                        <v-stepper-content step="3">
+                            <v-form ref="databaseSourcesForm">
+                                <v-container fluid>
+                                    <v-row>
+                                        <v-col cols="6">
+                                            <v-select
+                                                dense
+                                                label="Database sources"
+                                                hint="Select all database sources that should be filtered."
+                                                v-model="selectedSources"
+                                                :rules="[
+                                                            selectedSources.length > 0 ||
+                                                            'At least one source should be selected'
+                                                        ]"
+                                                persistent-hint
+                                                multiple
+                                                :items="sources">
+                                            </v-select>
+                                        </v-col>
+                                        <v-col cols="6">
+                                            <div>
+                                                <span class="font-weight-bold">Note:</span> the most recent UniProtKB
+                                                version will be used for the construction of this database. The most recent
+                                                version currently is {{ selectedVersion }}.
+                                            </div>
+    <!--                                        <v-select-->
+    <!--                                        dense-->
+    <!--                                        label="Database version"-->
+    <!--                                        :items="versions"-->
+    <!--                                        v-model="selectedVersion"-->
+    <!--                                        :rules="[value => !! value || 'You must select a UniProtKB source']"-->
+    <!--                                        hint="Select the version of the UniProtKB source that should be processed."-->
+    <!--                                        persistent-hint>-->
+    <!--                                        </v-select>-->
+                                        </v-col>
+                                    </v-row>
+                                    <v-row>
+                                        <v-col cols="12">
+                                            <v-btn class="mr-1" @click="currentStep--">Go back</v-btn>
+                                            <v-btn color="primary" @click="validateDatabaseSourcesAndContinue()">
+                                                Continue
+                                            </v-btn>
+                                        </v-col>
+                                    </v-row>
+                                </v-container>
+                            </v-form>
+                        </v-stepper-content>
+
+                        <v-stepper-step step="4" :complete="currentStep > 4">
+                            Filter
+                            <small>Select which organisms will be present in the output database</small>
+                        </v-stepper-step>
+
+                        <v-stepper-content step="4">
                             <v-container fluid>
                                 <v-row>
-                                    <v-col cols="6">
-                                        <v-select
-                                            dense
-                                            label="Database sources"
-                                            hint="Select all database sources that should be filtered."
-                                            v-model="selectedSources"
-                                            :rules="[
-                                                    selectedSources.length > 0 ||
-                                                    'At least one source should be selected'
-                                                ]"
-                                            persistent-hint
-                                            multiple
-                                            :items="sources">
-                                        </v-select>
-                                    </v-col>
-                                    <v-col cols="6">
+                                    <v-col cols="12">
                                         <div>
-                                            <span class="font-weight-bold">Note:</span> the most recent UniProtKB
-                                            version will be used for the construction of this database. The most recent
-                                            version currently is {{ selectedVersion }}.
+                                            <taxa-browser
+                                                v-model="selectedTaxa"
+                                                :swissprot-selected="selectedSources.includes('SwissProt')"
+                                                :trembl-selected="selectedSources.includes('TrEMBL')"
+                                            />
                                         </div>
-<!--                                        <v-select-->
-<!--                                            dense-->
-<!--                                            label="Database version"-->
-<!--                                            :items="versions"-->
-<!--                                            v-model="selectedVersion"-->
-<!--                                            :rules="[value => !! value || 'You must select a UniProtKB source']"-->
-<!--                                            hint="Select the version of the UniProtKB source that should be processed."-->
-<!--                                            persistent-hint>-->
-<!--                                        </v-select>-->
                                     </v-col>
                                 </v-row>
                                 <v-row>
                                     <v-col cols="12">
-                                        <v-btn color="primary" @click="selectDatabaseSources">Continue</v-btn>
+                                        <v-btn class="mr-1" @click="currentStep--">Go back</v-btn>
+                                        <v-btn color="primary" @click="buildDatabase()">Build database</v-btn>
                                     </v-col>
                                 </v-row>
                             </v-container>
-                        </v-form>
-                    </v-stepper-content>
-
-                    <v-stepper-step step="3" :complete="currentStep > 3">
-                        Filter
-                        <small>Select which organisms will be present in the output database</small>
-                    </v-stepper-step>
-
-                    <v-stepper-content step="3">
-                        <v-container>
-                            <v-row>
-                                <v-col cols="12">
-                                    <div>
-                                        <taxa-browser
-                                            v-model="selectedTaxa"
-                                            :swissprot-selected="selectedSources.includes('SwissProt')"
-                                            :trembl-selected="selectedSources.includes('TrEMBL')"
-                                        />
-                                    </div>
-                                </v-col>
-                            </v-row>
-                            <v-row>
-                                <v-col cols="12">
-                                    <v-btn class="mr-2" @click="currentStep = 1">Go back</v-btn>
-                                    <v-btn color="primary" @click="buildDatabase()">Build database</v-btn>
-                                </v-col>
-                            </v-row>
-                        </v-container>
-                    </v-stepper-content>
+                        </v-stepper-content>
+                    </div>
                 </v-stepper>
             </v-card-text>
         </v-card>
@@ -183,6 +276,8 @@ import CachedNcbiResponseCommunicator from "@/logic/communication/taxonomic/ncbi
 import ConfigurationManager from "@/logic/configuration/ConfigurationManager";
 import CustomDatabase from "@/logic/custom_database/CustomDatabase";
 import ProteomeCommunicator from "@/logic/communication/proteomes/ProteomeCommunicator";
+import Proteome from "@/logic/communication/proteomes/Proteome";
+import UniProtConstants from "@/logic/communication/uniprot/UniProtConstants";
 
 @Component({
     components: { TaxaBrowser }
@@ -222,6 +317,11 @@ export default class CreateCustomDatabase extends Vue {
     private selectedTaxa: NcbiTaxon[] = [];
 
     private currentStep = 1;
+    private useProteomeFilter = false;
+
+    private referenceProteome = "";
+    private referenceProteomes: Proteome[] = [];
+    private referenceProteomeError: string = "";
 
     private async mounted() {
         this.onValueChanged();
@@ -274,9 +374,6 @@ export default class CreateCustomDatabase extends Vue {
 
             this.selectedVersion = lastVersion;
 
-            const results = await ProteomeCommunicator.getProteomeById("UP000464341");
-            console.log(results);
-
             this.loading = false;
         } catch (e) {
             console.error(e);
@@ -284,25 +381,52 @@ export default class CreateCustomDatabase extends Vue {
         }
     }
 
-    private manuallyFilterDatabase(): void {
+    private validateDatabaseNameAndContinue(): void {
         // @ts-ignore
-        if (this.$refs.databaseDetailsForm.validate()) {
-            // this.selectProteomes = false;
+        if (this.$refs.databaseNameForm.validate()) {
             this.currentStep = 2;
         }
     }
 
-    private filterByProteomes(): void {
+    private manuallyFilterDatabase(): void {
+        this.useProteomeFilter = false;
+        this.currentStep = 3;
+
+    }
 
+    private filterByProteomes(): void {
+        this.useProteomeFilter = true;
+        this.currentStep = 3;
     }
 
-    private selectDatabaseSources(): void {
+    private validateDatabaseSourcesAndContinue(): void {
         // @ts-ignore
         if (this.$refs.databaseSourcesForm.validate()) {
-            this.currentStep = 3;
+            this.currentStep = 4;
         }
     }
 
+    private async addReferenceProteome(): Promise<void> {
+        this.referenceProteomeError = "";
+        try {
+            const proteome = await ProteomeCommunicator.getProteomeById(this.referenceProteome);
+            if (proteome === null) {
+                this.referenceProteomeError = "This proteome does not exist.";
+            } else if (this.referenceProteomes.some(p => p.id === proteome.id)) {
+                this.referenceProteomeError = "This proteome is already added.";
+            } else {
+                this.referenceProteomes.push(proteome);
+                this.referenceProteome = "";
+            }
+        } catch (e) {
+            this.referenceProteomeError = e.message;
+        }
+    }
+
+    private removeReferenceProteome(idx: number): void {
+        this.referenceProteomes.splice(idx, 1);
+    }
+
     private async buildDatabase(): Promise<void> {
         // No filtering should be applied in this case (which means we pass only the root to the construction step of
         // the database).
@@ -317,6 +441,7 @@ export default class CreateCustomDatabase extends Vue {
             [
                 this.databaseName,
                 convertedSources,
+                convertedSources.map(s => UniProtConstants.SOURCE_URLS[s]),
                 this.selectedTaxa.map(taxon => taxon.id),
                 this.selectedVersion
             ]
@@ -327,6 +452,26 @@ export default class CreateCustomDatabase extends Vue {
         this.resetWizard();
     }
 
+    private async buildDatabaseFromReferenceProteomes(): Promise<void> {
+        this.$store.dispatch(
+            "customDatabases/buildDatabase",
+            [
+                this.databaseName,
+                this.referenceProteomes.map(p => p.id),
+                this.referenceProteomes.map(
+                    p => `https://rest.uniprot.org/uniprotkb/stream?compressed=true\\&format=xml\\&query=\\(\\(proteome:${p.id}\\)\\)`
+                ),
+                // No filtering should be applied in this case
+                [1],
+                this.selectedVersion
+            ]
+        );
+        this.dialogActive = false;
+
+        // After a database construction was started, we need to reset this wizard and prepare it for the next user.
+        this.resetWizard();
+    }
+
     /**
      * Reset the database construction wizard in this dialog to it's initial state.
      */
diff --git a/src/components/taxon/TaxaBrowser.vue b/src/components/taxon/TaxaBrowser.vue
index c59d06d..73da084 100644
--- a/src/components/taxon/TaxaBrowser.vue
+++ b/src/components/taxon/TaxaBrowser.vue
@@ -75,7 +75,7 @@
                     </v-row>
                     <v-row>
                         <div>
-                            <span v-if="uniprotRecordsLoading">Computing database size...</span>
+                            <span v-if="uniprotRecordsHelper.isExecuting()">Computing database size...</span>
                             <span v-else>
                                 Resulting database will contain {{ formattedUniprotRecords }} UniProtKB records.
                             </span>
@@ -198,6 +198,7 @@ import { Prop, Watch } from "vue-property-decorator";
 import { DataOptions } from "vuetify";
 import { promises as fs } from "fs";
 import MetadataCommunicator from "@/logic/communication/metadata/MetadataCommunicator";
+import AsyncHelper from "@/logic/AsyncHelper";
 
 const { dialog } = require("@electron/remote");
 
@@ -298,7 +299,7 @@ export default class TaxaBrowser extends Vue {
     private showSearchHintActive = false;
 
     private uniprotRecords = 0;
-    private uniprotRecordsLoading = false;
+    private uniprotRecordsHelper = new AsyncHelper<number>();
 
     private get formattedUniprotRecords(): string {
         return StringUtils.toHumanReadableNumber(this.uniprotRecords);
@@ -364,19 +365,24 @@ export default class TaxaBrowser extends Vue {
     }
 
     @Watch("selectedItems")
+    @Watch("swissprotSelected")
+    @Watch("tremblSelected")
     private onSelectedItemsChanged(): void {
         this.$emit("input", this.selectedItems);
         this.computeUniprotRecords();
     }
 
     private async computeUniprotRecords(): Promise<void> {
-        this.uniprotRecordsLoading = true;
-        this.uniprotRecords = await MetadataCommunicator.getUniProtRecordCount(
-            this.selectedItems.map(taxon => taxon.id),
-            this.swissprotSelected,
-            this.tremblSelected
+        this.uniprotRecordsHelper.performIfLast(
+            () => MetadataCommunicator.getUniProtRecordCount(
+                this.selectedItems.map(taxon => taxon.id),
+                this.swissprotSelected,
+                this.tremblSelected
+            ),
+            (count) => {
+                this.uniprotRecords = count;
+            }
         );
-        this.uniprotRecordsLoading = false;
     }
 
     private getRankColor(rank: string): string {
diff --git a/src/logic/AsyncHelper.ts b/src/logic/AsyncHelper.ts
new file mode 100644
index 0000000..b13741d
--- /dev/null
+++ b/src/logic/AsyncHelper.ts
@@ -0,0 +1,41 @@
+/**
+ * This class is a helper utility for async operations. It provides a method that can be used if multiple async
+ * requests can be made simultaneously, but only the last result should actually be used.
+ */
+export default class AsyncHelper<T> {
+    private asyncCount: number = 0;
+    private executing: boolean = false;
+
+    /**
+     * Use this method if a specific async operation could be executed more than once, but only the result of the final
+     * call should be retained and used. Note that the final call is defined as the one that was last called by this
+     * method, it is not necessarily the one that finishes last.
+     *
+     * @param asyncResult The function that can be executed multiple times and for which only the results of the final
+     * evocation should be retained.
+     * @param mutation A mutation that should be performed upon receiving the results of the async function. This will
+     * only be executed if the results of the final invocation of this method are available.
+     */
+    public async performIfLast(
+        asyncResult: () => Promise<T>,
+        mutation: (result: T) => void
+    ): Promise<void> {
+        this.executing = true;
+        this.asyncCount++;
+        const count = this.asyncCount;
+
+        const result = await asyncResult();
+
+        if (count === this.asyncCount) {
+            mutation(result);
+            this.executing = false;
+        }
+    }
+
+    /**
+     * Is this helper currently waiting for a call to finish or not?
+     */
+    public isExecuting(): boolean {
+        return this.executing;
+    }
+}
diff --git a/src/logic/communication/docker/DockerCommunicator.ts b/src/logic/communication/docker/DockerCommunicator.ts
index 6217ae5..c1f345a 100644
--- a/src/logic/communication/docker/DockerCommunicator.ts
+++ b/src/logic/communication/docker/DockerCommunicator.ts
@@ -27,8 +27,8 @@ export default class DockerCommunicator {
     public static readonly WEB_COMPONENT_PUBLIC_URL = "http://localhost";
     public static readonly WEB_COMPONENT_PUBLIC_PORT = "3000";
 
-    public static readonly UNIPEPT_DB_IMAGE_NAME = "ghcr.io/unipept/unipept-database:1.0";
-    public static readonly UNIPEPT_WEB_IMAGE_NAME = "ghcr.io/unipept/unipept-web:1.0";
+    public static readonly UNIPEPT_DB_IMAGE_NAME = "ghcr.io/unipept/unipept-database:1.1";
+    public static readonly UNIPEPT_WEB_IMAGE_NAME = "ghcr.io/unipept/unipept-web:1.1";
 
     public static connection: Dockerode;
 
diff --git a/src/logic/communication/uniprot/UniProtConstants.ts b/src/logic/communication/uniprot/UniProtConstants.ts
new file mode 100644
index 0000000..596ffdd
--- /dev/null
+++ b/src/logic/communication/uniprot/UniProtConstants.ts
@@ -0,0 +1,6 @@
+export default class UniProtConstants {
+    public static readonly SOURCE_URLS = {
+        "trembl": "https://ftp.expasy.org/databases/uniprot/current_release/knowledgebase/complete/uniprot_trembl.xml.gz",
+        "swissprot": "https://ftp.expasy.org/databases/uniprot/current_release/knowledgebase/complete/uniprot_sprot.xml.gz",
+    }
+}
diff --git a/src/logic/filesystem/analysis/AnalysisSourceManager.ts b/src/logic/filesystem/analysis/AnalysisSourceManager.ts
index 34daf9a..9515c45 100644
--- a/src/logic/filesystem/analysis/AnalysisSourceManager.ts
+++ b/src/logic/filesystem/analysis/AnalysisSourceManager.ts
@@ -7,6 +7,7 @@ import CustomDatabaseManager from "@/logic/filesystem/docker/CustomDatabaseManag
 import ConfigurationManager from "@/logic/configuration/ConfigurationManager";
 import CachedCustomDbAnalysisSource from "@/logic/communication/analysis/CachedCustomDbAnalysisSource";
 import { Store } from "vuex";
+import UniProtConstants from "@/logic/communication/uniprot/UniProtConstants";
 
 export default class AnalysisSourceManager {
     constructor(
@@ -64,6 +65,7 @@ export default class AnalysisSourceManager {
                         // TODO: we need to make sure here that a DB with this name does not already exist in the system
                         newDbName,
                         selectedSources,
+                        selectedSources.map(s => UniProtConstants.SOURCE_URLS[s]),
                         selectedTaxa,
                         sourceData.uniprot_version
                     ]
diff --git a/src/logic/filesystem/docker/CustomDatabaseManager.ts b/src/logic/filesystem/docker/CustomDatabaseManager.ts
index 957a86d..b2ad375 100644
--- a/src/logic/filesystem/docker/CustomDatabaseManager.ts
+++ b/src/logic/filesystem/docker/CustomDatabaseManager.ts
@@ -163,7 +163,9 @@ export default class CustomDatabaseManager {
         const dbs = await this.listAllDatabases(dbRootFolder);
 
         const possibleDbs = dbs.filter((db: CustomDatabase) => {
+            console.log("Checking: " + db.name);
             if (db.databaseVersion !== uniprotVersion) {
+                console.log("DB Version is different!");
                 return false;
             }
 
@@ -173,9 +175,13 @@ export default class CustomDatabaseManager {
                     selectedSources.sort()
                 )
             ) {
+                console.log(db.sourceTypes.sort());
+                console.log(selectedSources.sort());
+                console.log("Source types are different!");
                 return false;
             }
 
+            console.log("Checking selected taxa...");
             return Utils.compareAssays<NcbiId>(db.taxa.sort(), selectedTaxa.sort());
         });
 
diff --git a/src/state/CustomDatabaseStoreFactory.ts b/src/state/CustomDatabaseStoreFactory.ts
index fa8f885..7e848fc 100644
--- a/src/state/CustomDatabaseStoreFactory.ts
+++ b/src/state/CustomDatabaseStoreFactory.ts
@@ -42,28 +42,17 @@ export default class CustomDatabaseStoreFactory {
                 [
                     dbName,
                     databaseTypes,
+                    databaseSources,
                     taxa,
                     dbVersion
                 ]: [
                     string,
                     string[],
+                    string[],
                     NcbiId[],
                     string
                 ]
             ) {
-                const sourceUrlMap = new Map<string, string>([
-                    [
-                        "trembl",
-                        "https://ftp.expasy.org/databases/uniprot/current_release/knowledgebase/complete/uniprot_trembl.xml.gz"
-                    ],
-                    [
-                        "swissprot",
-                        "https://ftp.expasy.org/databases/uniprot/current_release/knowledgebase/complete/uniprot_sprot.xml.gz"
-                    ]
-                ]);
-
-                const databaseSources = databaseTypes.map(t => sourceUrlMap.get(t));
-
                 const db = new CustomDatabase(
                     dbName,
                     databaseSources,
@@ -205,16 +194,21 @@ export default class CustomDatabaseStoreFactory {
                     [
                         dbName,
                         databaseTypes,
+                        databaseSources,
                         taxa,
                         uniprotVersion
                     ]: [
                         string,
                         string[],
+                        string[],
                         NcbiId[],
                         string
                     ]
                 ) => {
-                    await this.buildDatabase(store, [dbName, databaseTypes, taxa, uniprotVersion]);
+                    await this.buildDatabase(
+                        store,
+                        [dbName, databaseTypes, databaseSources, taxa, uniprotVersion]
+                    );
                 }
             },
 
@@ -416,11 +410,13 @@ export default class CustomDatabaseStoreFactory {
         [
             dbName,
             databaseTypes,
+            databaseSources,
             taxa,
             uniprotVersion
         ]: [
             string,
             string[],
+            string[],
             NcbiId[],
             string
         ]
@@ -428,6 +424,7 @@ export default class CustomDatabaseStoreFactory {
         store.commit("CUSTOM_DB_ADD_DATABASE", [
             dbName,
             databaseTypes,
+            databaseSources,
             taxa,
             uniprotVersion
         ]);

From 387ca601a1a6f73a951c0be49b714f6707efb27f Mon Sep 17 00:00:00 2001
From: Pieter Verschaffelt <pieter.verschaffelt@ugent.be>
Date: Fri, 7 Oct 2022 10:28:13 +0200
Subject: [PATCH 3/6] update database schema for storing referene proteomes

---
 src/db/migrations/v5_to_v6.sql                | 21 +++++
 src/db/schemas/schema_v6.sql                  | 76 +++++++++++++++++++
 .../analysis/AnalysisSourceManager.ts         | 33 +++-----
 .../assay/AssayFileSystemDataReader.ts        |  2 +-
 .../filesystem/database/DatabaseManager.ts    |  8 +-
 .../database/DatabaseMigratorV5ToV6.ts        | 35 +++++++++
 src/logic/filesystem/database/Schema.ts       |  9 +--
 .../docker/CustomDatabaseManager.ts           |  6 --
 8 files changed, 153 insertions(+), 37 deletions(-)
 create mode 100644 src/db/migrations/v5_to_v6.sql
 create mode 100644 src/db/schemas/schema_v6.sql
 create mode 100644 src/logic/filesystem/database/DatabaseMigratorV5ToV6.ts

diff --git a/src/db/migrations/v5_to_v6.sql b/src/db/migrations/v5_to_v6.sql
new file mode 100644
index 0000000..2f57c8f
--- /dev/null
+++ b/src/db/migrations/v5_to_v6.sql
@@ -0,0 +1,21 @@
+DROP TABLE analysis_source;
+CREATE TABLE analysis_source (
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    type TEXT CHECK( type in ('online', 'custom_db') ) NOT NULL,
+    /**
+    * If the type of the AnalysisSource is online, this field indicates which API endpoint was used for the analysis.
+    */
+    endpoint TEXT,
+    /* UniProt version that was used to process the underlying data (e.g. 2022.02). */
+    uniprot_version TEXT NOT NULL,
+    /**
+    * Comma-delimited list of NCBI taxon ID's that are used for filtering in this database. This field will only be
+    * used if the type of this analysis source is "custom_db".
+    */
+    selected_taxa TEXT,
+    /**
+    * Comma-delimited list of databases (or database sources) that are used to construct this database. This list
+    * could include things like swissprot, trembl or reference proteome identifiers.
+    */
+    sources TEXT
+);
diff --git a/src/db/schemas/schema_v6.sql b/src/db/schemas/schema_v6.sql
new file mode 100644
index 0000000..29d2e1a
--- /dev/null
+++ b/src/db/schemas/schema_v6.sql
@@ -0,0 +1,76 @@
+CREATE TABLE studies (
+    id TEXT PRIMARY KEY,
+    name TEXT NOT NULL
+);
+
+CREATE TABLE assays (
+    id TEXT PRIMARY KEY,
+    name TEXT NOT NULL,
+    study_id TEXT NOT NULL,
+    configuration_id INT NOT NULL,
+    /*
+      AnalysisSource that was last selected for this assay. This endpoint does not necessarily need to be the same as the one
+      selected for the storage_metadata table. This endpoint is only used to detect whether the assay needs to be
+      recomputed or not.
+     */
+    analysis_source_id INTEGER,
+    FOREIGN KEY(study_id) REFERENCES studies(id),
+    FOREIGN KEY(configuration_id) REFERENCES search_configuration(id),
+    FOREIGN KEY(analysis_source_id) REFERENCES analysis_source(id)
+);
+
+CREATE TABLE search_configuration (
+    id INTEGER PRIMARY KEY,
+    equate_il INT NOT NULL,
+    filter_duplicates INT NOT NULL,
+    missing_cleavage_handling INT NOT NULL
+);
+
+CREATE TABLE peptide_trust (
+    assay_id TEXT NOT NULL,
+    missed_peptides TEXT NOT NULL,
+    matched_peptides INT NOT NULL,
+    searched_peptides INT NOT NULL,
+    PRIMARY KEY(assay_id)
+);
+
+CREATE TABLE storage_metadata (
+    assay_id TEXT NOT NULL,
+    configuration_id INT NOT NULL,
+    /*
+       Hash of the files that are stored on the local filesystem. This hash can be used to verify the integrity of the
+       files containing the offline result data on the filesystem. Value for this column is the concatenation of
+       the hash for both the data buffer and index buffer files.
+    */
+    data_hash TEXT,
+    analysis_date TEXT,
+    analysis_source_id INTEGER,
+    PRIMARY KEY(assay_id),
+    FOREIGN KEY(configuration_id) REFERENCES search_configuration(id),
+    FOREIGN KEY(analysis_source_id) REFERENCES analysis_source(id)
+);
+
+CREATE TABLE analysis_source (
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    type TEXT CHECK( type in ('online', 'custom_db') ) NOT NULL,
+    /**
+     * If the type of the AnalysisSource is online, this field indicates which API endpoint was used for the analysis.
+     */
+    endpoint TEXT,
+    /* UniProt version that was used to process the underlying data (e.g. 2022.02). */
+    uniprot_version TEXT NOT NULL,
+    /**
+      * Comma-delimited list of NCBI taxon ID's that are used for filtering in this database. This field will only be
+      * used if the type of this analysis source is "custom_db".
+      */
+    selected_taxa TEXT,
+    /**
+      * Comma-delimited list of databases (or database sources) that are used to construct this database. This list
+      * could include things like swissprot, trembl or reference proteome identifiers.
+      */
+    sources TEXT
+);
+
+CREATE TABLE database_metadata (
+    application_version TEXT NOT NULL
+);
diff --git a/src/logic/filesystem/analysis/AnalysisSourceManager.ts b/src/logic/filesystem/analysis/AnalysisSourceManager.ts
index 9515c45..1266597 100644
--- a/src/logic/filesystem/analysis/AnalysisSourceManager.ts
+++ b/src/logic/filesystem/analysis/AnalysisSourceManager.ts
@@ -35,14 +35,7 @@ export default class AnalysisSourceManager {
         } else {
             const customDbManager = new CustomDatabaseManager();
 
-            const selectedSources: string[] = [];
-            if (sourceData.swissprot_selected === 1) {
-                selectedSources.push("swissprot");
-            }
-
-            if (sourceData.trembl_selected === 1) {
-                selectedSources.push("trembl");
-            }
+            const selectedSources: string[] = sourceData.sources.split(",");
 
             const configMng = new ConfigurationManager();
             const customDbStorageLocation: string = (await configMng.readConfiguration()).customDbStorageLocation;
@@ -107,11 +100,10 @@ export default class AnalysisSourceManager {
                             endpoint = ?,
                             uniprot_version = ?,
                             selected_taxa = ?,
-                            swissprot_selected = ?,
-                            trembl_selected = ?
+                            sources = ?
                         WHERE
                             id = ?
-                    `).run("online", source.endpoint, "N/A", "", 1, 1, sourceId);
+                    `).run("online", source.endpoint, "N/A", "", "", sourceId);
                 } else if (source instanceof CachedCustomDbAnalysisSource) {
                     const customDb = source.customDatabase;
                     db.prepare(`
@@ -121,8 +113,7 @@ export default class AnalysisSourceManager {
                             endpoint = ?,
                             uniprot_version = ?,
                             selected_taxa = ?,
-                            swissprot_selected = ?,
-                            trembl_selected = ?
+                            sources = ?
                         WHERE
                             id = ?
                     `).run(
@@ -130,8 +121,7 @@ export default class AnalysisSourceManager {
                         "",
                         customDb.databaseVersion,
                         customDb.taxa.join(","),
-                        customDb.sourceTypes.includes("swissprot") ? 1 : 0,
-                        customDb.sourceTypes.includes("trembl") ? 1 : 0,
+                        customDb.sourceTypes.join(","),
                         sourceId
                     );
                 }
@@ -144,23 +134,22 @@ export default class AnalysisSourceManager {
 
                 if (source instanceof OnlineAnalysisSource || source instanceof CachedOnlineAnalysisSource) {
                     runResult = db.prepare(`
-                        INSERT INTO analysis_source (type, endpoint, uniprot_version, swissprot_selected, trembl_selected) 
-                        VALUES (?, ?, ?, ?, ?)
-                    `).run("online", source.endpoint, "N/A", 1, 1);
+                        INSERT INTO analysis_source (type, endpoint, uniprot_version, sources) 
+                        VALUES (?, ?, ?, ?)
+                    `).run("online", source.endpoint, "N/A", "");
                 } else if (source instanceof CachedCustomDbAnalysisSource) {
                     const customDb = source.customDatabase;
 
                     runResult = db.prepare(`
                         INSERT INTO analysis_source (
-                            type, uniprot_version, selected_taxa, swissprot_selected, trembl_selected
+                            type, uniprot_version, selected_taxa, sources
                         )
-                        VALUES (?, ?, ?, ?, ?)
+                        VALUES (?, ?, ?, ?)
                     `).run(
                         "custom_db",
                         customDb.databaseVersion,
                         customDb.taxa.join(","),
-                        customDb.sourceTypes.includes("swissprot") ? 1 : 0,
-                        customDb.sourceTypes.includes("trembl") ? 1 : 0
+                        customDb.sourceTypes.join(",")
                     );
                 }
 
diff --git a/src/logic/filesystem/assay/AssayFileSystemDataReader.ts b/src/logic/filesystem/assay/AssayFileSystemDataReader.ts
index 817a5db..273cf65 100644
--- a/src/logic/filesystem/assay/AssayFileSystemDataReader.ts
+++ b/src/logic/filesystem/assay/AssayFileSystemDataReader.ts
@@ -39,7 +39,7 @@ export default class AssayFileSystemDataReader extends FileSystemAssayVisitor {
     public async visitProteomicsAssay(mpAssay: ProteomicsAssay): Promise<void> {
         const path = `${this.directoryPath}${mpAssay.getName()}.pep`;
 
-        // First, try to read in all of the peptides for this assay.
+        // First, try to read in all the peptides for this assay.
         const peptidesString: string = await fs.readFile(path, {
             encoding: "utf-8"
         });
diff --git a/src/logic/filesystem/database/DatabaseManager.ts b/src/logic/filesystem/database/DatabaseManager.ts
index b747a09..cb7da58 100644
--- a/src/logic/filesystem/database/DatabaseManager.ts
+++ b/src/logic/filesystem/database/DatabaseManager.ts
@@ -1,5 +1,5 @@
 import Database, { Database as DbType } from "better-sqlite3";
-import async, { AsyncQueue } from "async";
+import async, { AsyncQueue, QueueObject } from "async";
 import DatabaseMigrator from "@/logic/filesystem/database/DatabaseMigrator";
 import DatabaseMigratorV0ToV1 from "@/logic/filesystem/database/DatabaseMigratorV0ToV1";
 import Schema from "@/logic/filesystem/database/Schema";
@@ -11,6 +11,7 @@ import path from "path";
 import DatabaseMigratorV2ToV3 from "@/logic/filesystem/database/DatabaseMigratorV2ToV3";
 import DatabaseMigratorV3ToV4 from "@/logic/filesystem/database/DatabaseMigratorV3ToV4";
 import DatabaseMigratorV4ToV5 from "@/logic/filesystem/database/DatabaseMigratorV4ToV5";
+import DatabaseMigratorV5ToV6 from "@/logic/filesystem/database/DatabaseMigratorV5ToV6";
 
 export default class DatabaseManager {
     // Reading and writing large assays to and from the database can easily take longer than 5 seconds, causing
@@ -19,7 +20,7 @@ export default class DatabaseManager {
     public static readonly DB_TIMEOUT: number = 15000;
 
     private db: DbType;
-    private queue: AsyncQueue<any>;
+    private queue: QueueObject<any>;
     // Version of the application that was last used with this database.
     private dbApplicationVersion: string;
 
@@ -32,7 +33,8 @@ export default class DatabaseManager {
         () => new DatabaseMigratorV1ToV2(path.dirname(this.dbLocation)),
         () => new DatabaseMigratorV2ToV3(path.dirname(this.dbLocation)),
         () => new DatabaseMigratorV3ToV4(path.dirname(this.dbLocation)),
-        () => new DatabaseMigratorV4ToV5(path.dirname(this.dbLocation))
+        () => new DatabaseMigratorV4ToV5(path.dirname(this.dbLocation)),
+        () => new DatabaseMigratorV5ToV6(path.dirname(this.dbLocation))
     ];
 
     /**
diff --git a/src/logic/filesystem/database/DatabaseMigratorV5ToV6.ts b/src/logic/filesystem/database/DatabaseMigratorV5ToV6.ts
new file mode 100644
index 0000000..ffe22e8
--- /dev/null
+++ b/src/logic/filesystem/database/DatabaseMigratorV5ToV6.ts
@@ -0,0 +1,35 @@
+import DatabaseMigrator from "@/logic/filesystem/database/DatabaseMigrator";
+import Database from "better-sqlite3";
+import v5_to_v6 from "raw-loader!@/db/migrations/v5_to_v6.sql";
+
+export default class DatabaseMigratorV5ToV6 implements DatabaseMigrator {
+    constructor(private readonly projectLocation: string) {}
+
+    public async upgrade(database: Database.Database): Promise<void> {
+        // First store all the information that's currently stored in the `analysis_source` table in memory.
+        const analysisSourceData = database.prepare("SELECT * FROM analysis_source").all();
+
+        database.pragma("foreign_keys = OFF");
+        // Then, update the schema of the database itself.
+        database.exec(v5_to_v6);
+
+        // Now, insert the old data in the newest database schema again.
+        for (const analysisSource of analysisSourceData) {
+            const sources = [];
+            if (analysisSource.swissprot_selected) {
+                sources.push("swissprot");
+            }
+
+            if (analysisSource.trembl_selected) {
+                sources.push("trembl");
+            }
+
+            database.prepare(`
+                INSERT INTO analysis_source (id, type, endpoint, uniprot_version, selected_taxa, sources)
+                VALUES (?, ?, ?, ?, ?, ?)
+            `).run(analysisSource.id, analysisSource.type, analysisSource.endpoint, analysisSource.uniprot_version, analysisSource.selected_taxa, sources.join(","));
+        }
+
+        database.pragma("foreign_keys = ON");
+    }
+}
diff --git a/src/logic/filesystem/database/Schema.ts b/src/logic/filesystem/database/Schema.ts
index 600068b..ac58cea 100644
--- a/src/logic/filesystem/database/Schema.ts
+++ b/src/logic/filesystem/database/Schema.ts
@@ -1,8 +1,8 @@
-import schema_v5 from "raw-loader!@/db/schemas/schema_v5.sql";
+import schema_v6 from "raw-loader!@/db/schemas/schema_v6.sql";
 
 export default class Schema {
-    public static LATEST_VERSION = 5;
-    public static LATEST_SCHEMA: string = schema_v5;
+    public static LATEST_VERSION = 6;
+    public static LATEST_SCHEMA: string = schema_v6;
 }
 
 // Type definitions for all the different rows in the database schema.
@@ -48,8 +48,7 @@ export type AnalysisSourceTableRow = {
     endpoint: string,
     uniprot_version: string,
     selected_taxa: string,
-    swissprot_selected: number,
-    trembl_selected: number
+    sources: string
 }
 
 export type DatabaseMetadataTableRow = {
diff --git a/src/logic/filesystem/docker/CustomDatabaseManager.ts b/src/logic/filesystem/docker/CustomDatabaseManager.ts
index b2ad375..957a86d 100644
--- a/src/logic/filesystem/docker/CustomDatabaseManager.ts
+++ b/src/logic/filesystem/docker/CustomDatabaseManager.ts
@@ -163,9 +163,7 @@ export default class CustomDatabaseManager {
         const dbs = await this.listAllDatabases(dbRootFolder);
 
         const possibleDbs = dbs.filter((db: CustomDatabase) => {
-            console.log("Checking: " + db.name);
             if (db.databaseVersion !== uniprotVersion) {
-                console.log("DB Version is different!");
                 return false;
             }
 
@@ -175,13 +173,9 @@ export default class CustomDatabaseManager {
                     selectedSources.sort()
                 )
             ) {
-                console.log(db.sourceTypes.sort());
-                console.log(selectedSources.sort());
-                console.log("Source types are different!");
                 return false;
             }
 
-            console.log("Checking selected taxa...");
             return Utils.compareAssays<NcbiId>(db.taxa.sort(), selectedTaxa.sort());
         });
 

From cf129b1adae2e4a9566ab85b3446c23001488463 Mon Sep 17 00:00:00 2001
From: Pieter Verschaffelt <pieter.verschaffelt@ugent.be>
Date: Fri, 7 Oct 2022 14:27:12 +0200
Subject: [PATCH 4/6] Fixed resetting create database form after creation

---
 .../custom-database/CreateCustomDatabase.vue  | 45 +++++++++++++------
 1 file changed, 32 insertions(+), 13 deletions(-)

diff --git a/src/components/custom-database/CreateCustomDatabase.vue b/src/components/custom-database/CreateCustomDatabase.vue
index d3d0806..b75fc3e 100644
--- a/src/components/custom-database/CreateCustomDatabase.vue
+++ b/src/components/custom-database/CreateCustomDatabase.vue
@@ -110,17 +110,26 @@
                         <v-stepper-content step="3">
                             <v-container fluid>
                                 <v-row>
-                                    <v-col cols="12" class="d-flex">
-                                        <v-text-field
-                                            dense
-                                            label="Reference proteome identifier"
-                                            hint="Provide a unique UniProtKB reference proteome identifier (e.g. UP000005640)"
-                                            persistent-hint
-                                            :error="referenceProteomeError !== ''"
-                                            :error-messages="referenceProteomeError"
-                                            v-model="referenceProteome">
-                                        </v-text-field>
-                                        <v-btn color="primary ml-2" @click="addReferenceProteome()">Add</v-btn>
+                                    <v-col cols="12">
+                                        <div class="d-flex">
+                                            <v-text-field
+                                                dense
+                                                label="Reference proteome identifier"
+                                                hint="Provide a unique UniProtKB reference proteome identifier (e.g. UP000005640)."
+                                                persistent-hint
+                                                :error="referenceProteomeError !== ''"
+                                                :error-messages="referenceProteomeError"
+                                                v-model="referenceProteome">
+                                            </v-text-field>
+                                            <v-btn color="primary ml-2" @click="addReferenceProteome()">Add</v-btn>
+                                        </div>
+                                        <div>
+                                            <span class="font-weight-bold">Hint:</span> browse
+                                            <a @click="openInBrowser('https://www.uniprot.org/proteomes')">
+                                                https://www.uniprot.org/proteomes
+                                            </a>
+                                            for a list of all available reference proteomes.
+                                        </div>
                                     </v-col>
                                 </v-row>
                                 <v-row>
@@ -270,7 +279,7 @@ import axios from "axios";
 
 import https from "https";
 import TaxaBrowser from "@/components/taxon/TaxaBrowser.vue";
-import { NcbiTaxon } from "unipept-web-components";
+import { NcbiTaxon, NetworkUtils } from "unipept-web-components";
 import { Prop, Watch } from "vue-property-decorator";
 import CachedNcbiResponseCommunicator from "@/logic/communication/taxonomic/ncbi/CachedNcbiResponseCommunicator";
 import ConfigurationManager from "@/logic/configuration/ConfigurationManager";
@@ -329,6 +338,10 @@ export default class CreateCustomDatabase extends Vue {
         await this.retrieveUniProtVersions();
     }
 
+    private openInBrowser(url: string) {
+        NetworkUtils.openInBrowser(url);
+    }
+
     /**
      * Checks if the provided database name is unique (i.e. does not already exists). Databases with the same name
      * cannot exist at the same time due to compatibility issues.
@@ -479,11 +492,17 @@ export default class CreateCustomDatabase extends Vue {
         this.currentStep = 1;
         this.selectedTaxa.splice(0, this.selectedTaxa.length);
         this.databaseName = "";
-        (this.$refs.databaseForm as any).reset();
+        (this.$refs.databaseNameForm as any).reset();
+        if (this.$refs.databaseSourcesForm) {
+            (this.$refs.databaseSourcesForm as any).reset();
+        }
         this.selectedSources.splice(0, this.selectedSources.length);
         this.selectedSources.push(...this.sources);
         this.selectedVersion = this.sources[0];
         this.selectedMirror = "EU (Expasy)";
+        this.referenceProteome = "";
+        this.referenceProteomes.splice(0, this.referenceProteomes.length);
+        this.referenceProteomeError = "";
         this.error = false;
     }
 

From 6775586771853ff59ec9e72e9b5d2f716d4e9648 Mon Sep 17 00:00:00 2001
From: Pieter Verschaffelt <pieter.verschaffelt@ugent.be>
Date: Fri, 7 Oct 2022 15:08:38 +0200
Subject: [PATCH 5/6] Duplicating custom DB with reference proteomes works now

---
 .../custom-database/CreateCustomDatabase.vue  | 35 +++++++++++++++++--
 src/components/pages/CustomDatabasePage.vue   |  2 +-
 src/components/proteome/ProteomeBrowser.vue   | 17 ---------
 3 files changed, 34 insertions(+), 20 deletions(-)
 delete mode 100644 src/components/proteome/ProteomeBrowser.vue

diff --git a/src/components/custom-database/CreateCustomDatabase.vue b/src/components/custom-database/CreateCustomDatabase.vue
index b75fc3e..6819e51 100644
--- a/src/components/custom-database/CreateCustomDatabase.vue
+++ b/src/components/custom-database/CreateCustomDatabase.vue
@@ -524,19 +524,50 @@ export default class CreateCustomDatabase extends Vue {
     }
 
     @Watch("value")
-    private onValueChanged() {
+    private async onValueChanged(): Promise<void> {
         if (this.value) {
             // Reset to the default supplied values.
+            if (this.databaseNameDefault !== "") {
+                if (
+                    this.selectedSourcesDefault.includes("trembl") ||
+                    this.selectedSourcesDefault.includes("swissprot")
+                ) {
+                    this.useProteomeFilter = false;
+                } else {
+                    this.referenceProteomes.splice(0, this.referenceProteomes.length);
+                    for (const proteomeId of this.selectedSourcesDefault) {
+                        this.referenceProteomes.push(await ProteomeCommunicator.getProteomeById(proteomeId));
+                    }
+
+                    this.useProteomeFilter = true;
+                }
+            }
+
             this.selectedSources.splice(0, this.selectedSources.length);
             this.selectedSources.push(...this.selectedSourcesDefault.map(
                 (x: string) => {
-                    return { "trembl": "TrEMBL", "swissprot": "SwissProt" }[x]
+                    const translationTable = { "trembl": "TrEMBL", "swissprot": "SwissProt" };
+                    if (x in translationTable) {
+                        return translationTable[x];
+                    } else {
+                        return x;
+                    }
                 }
             ));
             this.databaseName = this.databaseNameDefault;
             // this.selectedVersion = this.selectedVersionDefault;
             this.selectedTaxa.splice(0, this.selectedTaxa.length);
             this.selectedTaxa.push(...this.selectedTaxaDefault);
+
+            if (
+                this.selectedSourcesDefault.includes("trembl") ||
+                this.selectedSourcesDefault.includes("swissprot")
+            ) {
+                await this.$nextTick();
+                this.currentStep = 4;
+            } else {
+                this.currentStep = 3;
+            }
         }
         this.dialogActive = this.value;
     }
diff --git a/src/components/pages/CustomDatabasePage.vue b/src/components/pages/CustomDatabasePage.vue
index 51f66e8..1f25906 100644
--- a/src/components/pages/CustomDatabasePage.vue
+++ b/src/components/pages/CustomDatabasePage.vue
@@ -430,7 +430,7 @@ export default class CustomDatabasePage extends Vue {
         this.selectedTaxaDefault.splice(0, this.selectedTaxaDefault.length);
         const ncbiOntologyProcessor = new NcbiOntologyProcessor(new CachedNcbiResponseCommunicator());
         const ontology = await ncbiOntologyProcessor.getOntologyByIds(db.taxa);
-        this.selectedTaxaDefault.push(...(db.taxa.map(id => ontology.getDefinition(id))));
+        this.selectedTaxaDefault.push(...(db.taxa.filter(id => id !== 1).map(id => ontology.getDefinition(id))));
 
         this.createDatabaseDialog = true;
     }
diff --git a/src/components/proteome/ProteomeBrowser.vue b/src/components/proteome/ProteomeBrowser.vue
deleted file mode 100644
index 524f73e..0000000
--- a/src/components/proteome/ProteomeBrowser.vue
+++ /dev/null
@@ -1,17 +0,0 @@
-<template>
-
-</template>
-
-<script lang="ts">
-import Vue from "vue";
-import Component from "vue-class-component";
-
-@Component
-export default class ProteomeBrowser extends Vue {
-
-}
-</script>
-
-<style scoped>
-
-</style>

From 3c0e1f00d356f406158456e309fd023a48f39815 Mon Sep 17 00:00:00 2001
From: Pieter Verschaffelt <pieter.verschaffelt@ugent.be>
Date: Fri, 7 Oct 2022 15:18:20 +0200
Subject: [PATCH 6/6] Add reference proteome total protein count

---
 .../custom-database/CreateCustomDatabase.vue  | 108 ++++++++++--------
 1 file changed, 62 insertions(+), 46 deletions(-)

diff --git a/src/components/custom-database/CreateCustomDatabase.vue b/src/components/custom-database/CreateCustomDatabase.vue
index 6819e51..d0cde51 100644
--- a/src/components/custom-database/CreateCustomDatabase.vue
+++ b/src/components/custom-database/CreateCustomDatabase.vue
@@ -134,39 +134,45 @@
                                 </v-row>
                                 <v-row>
                                     <v-col cols="12">
-                                        <v-simple-table v-if="referenceProteomes.length > 0">
-                                            <template v-slot:default>
-                                                <thead>
-                                                <tr>
-                                                    <th class="text-left">Proteome ID</th>
-                                                    <th class="text-left">Organism name</th>
-                                                    <th class="text-left">Protein count</th>
-                                                    <th class="text-center">Actions</th>
-                                                </tr>
-                                                </thead>
-                                                <tbody>
-                                                <tr v-for="(proteome, index) in referenceProteomes" :key="index">
-                                                    <td>{{ proteome.id }}</td>
-                                                    <td>{{ proteome.organismName }}</td>
-                                                    <td>{{ proteome.proteinCount }}</td>
-                                                    <td class="text-center">
-                                                        <v-tooltip bottom open-delay="500">
-                                                            <template v-slot:activator="{ on }">
-                                                                <v-btn
-                                                                    icon
-                                                                    @click="removeReferenceProteome(index)"
-                                                                    v-on="on"
-                                                                    color="red">
-                                                                    <v-icon>mdi-delete</v-icon>
-                                                                </v-btn>
-                                                            </template>
-                                                            <span>Remove reference proteome from selection</span>
-                                                        </v-tooltip>
-                                                    </td>
-                                                </tr>
-                                                </tbody>
-                                            </template>
-                                        </v-simple-table>
+                                        <div v-if="referenceProteomes.length > 0">
+                                            <v-simple-table>
+                                                <template v-slot:default>
+                                                    <thead>
+                                                        <tr>
+                                                            <th class="text-left">Proteome ID</th>
+                                                            <th class="text-left">Organism name</th>
+                                                            <th class="text-left">Protein count</th>
+                                                            <th class="text-center">Actions</th>
+                                                        </tr>
+                                                    </thead>
+                                                    <tbody>
+                                                        <tr v-for="(proteome, index) in referenceProteomes" :key="index">
+                                                            <td>{{ proteome.id }}</td>
+                                                            <td>{{ proteome.organismName }}</td>
+                                                            <td>{{ proteome.proteinCount }}</td>
+                                                            <td class="text-center">
+                                                                <v-tooltip bottom open-delay="500">
+                                                                    <template v-slot:activator="{ on }">
+                                                                        <v-btn
+                                                                            icon
+                                                                            @click="removeReferenceProteome(index)"
+                                                                            v-on="on"
+                                                                            color="red">
+                                                                            <v-icon>mdi-delete</v-icon>
+                                                                        </v-btn>
+                                                                    </template>
+                                                                    <span>Remove reference proteome from selection</span>
+                                                                </v-tooltip>
+                                                            </td>
+                                                        </tr>
+                                                    </tbody>
+                                                </template>
+                                            </v-simple-table>
+                                            <span>
+                                                Resulting database will contain {{ totalReferenceProteins }} UniProtKB
+                                                records.
+                                            </span>
+                                        </div>
                                         <div v-else>
                                             No reference proteomes have been added yet. Find one in the search bar above
                                             and click the "Add" button.
@@ -332,6 +338,10 @@ export default class CreateCustomDatabase extends Vue {
     private referenceProteomes: Proteome[] = [];
     private referenceProteomeError: string = "";
 
+    get totalReferenceProteins(): number {
+        return this.referenceProteomes.reduce((acc, proteome) => acc + proteome.proteinCount, 0);
+    }
+
     private async mounted() {
         this.onValueChanged();
         this.selectedMirror = this.getMostSuitableMirror();
@@ -492,7 +502,9 @@ export default class CreateCustomDatabase extends Vue {
         this.currentStep = 1;
         this.selectedTaxa.splice(0, this.selectedTaxa.length);
         this.databaseName = "";
-        (this.$refs.databaseNameForm as any).reset();
+        if (this.$refs.databaseNameForm) {
+            (this.$refs.databaseNameForm as any).reset();
+        }
         if (this.$refs.databaseSourcesForm) {
             (this.$refs.databaseSourcesForm as any).reset();
         }
@@ -527,20 +539,24 @@ export default class CreateCustomDatabase extends Vue {
     private async onValueChanged(): Promise<void> {
         if (this.value) {
             // Reset to the default supplied values.
-            if (this.databaseNameDefault !== "") {
-                if (
-                    this.selectedSourcesDefault.includes("trembl") ||
-                    this.selectedSourcesDefault.includes("swissprot")
-                ) {
-                    this.useProteomeFilter = false;
-                } else {
-                    this.referenceProteomes.splice(0, this.referenceProteomes.length);
-                    for (const proteomeId of this.selectedSourcesDefault) {
-                        this.referenceProteomes.push(await ProteomeCommunicator.getProteomeById(proteomeId));
-                    }
+            if (this.databaseNameDefault === "") {
+                this.resetWizard();
+                this.dialogActive = this.value;
+                return;
+            }
 
-                    this.useProteomeFilter = true;
+            if (
+                this.selectedSourcesDefault.includes("trembl") ||
+                this.selectedSourcesDefault.includes("swissprot")
+            ) {
+                this.useProteomeFilter = false;
+            } else {
+                this.referenceProteomes.splice(0, this.referenceProteomes.length);
+                for (const proteomeId of this.selectedSourcesDefault) {
+                    this.referenceProteomes.push(await ProteomeCommunicator.getProteomeById(proteomeId));
                 }
+
+                this.useProteomeFilter = true;
             }
 
             this.selectedSources.splice(0, this.selectedSources.length);