From 8d228cd1b02bd6ebc009591fa05635a69c674bbf Mon Sep 17 00:00:00 2001
From: Matthias Bernt <m.bernt@ufz.de>
Date: Thu, 11 Apr 2024 15:46:43 +0200
Subject: [PATCH] make suite, finish tools

- add assumptions of meryldb zip files
- output names
---
 tools/meryl/.shed.yml            | 16 ++++++++++++----
 tools/meryl/arithmetic-kmers.xml | 11 ++++++++---
 tools/meryl/count-kmers.xml      |  6 ++++--
 tools/meryl/filter-kmers.xml     |  8 ++++++--
 tools/meryl/groups-kmers.xml     | 14 ++++++++++++--
 tools/meryl/histogram-kmers.xml  |  9 ++++++---
 tools/meryl/macros.xml           | 13 +++++++++++++
 tools/meryl/print.xml            |  3 +--
 tools/meryl/trio-mode.xml        | 21 +++++++++------------
 9 files changed, 71 insertions(+), 30 deletions(-)

diff --git a/tools/meryl/.shed.yml b/tools/meryl/.shed.yml
index 92ccb006234..87b8c9256fe 100644
--- a/tools/meryl/.shed.yml
+++ b/tools/meryl/.shed.yml
@@ -1,10 +1,18 @@
-categories:
-- Assembly
+name: meryl
+owner: iuc
 description: Meryl a k-mer counter.
 long_description: |
    Meryl a k-mer counter. It is built into the Celera assembler and is also available as a stand-alone application. Meryl uses a sorting-based approach that sorts k-mers in lexicographical order.
-name: meryl
-owner: iuc
 remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/master/tools/meryl
 homepage_url: https://github.com/marbl/meryl
 type: unrestricted
+categories:
+- Assembly
+auto_tool_repositories:
+  name_template: "{{ tool_id }}"
+  description_template: "Wrapper for Meryl function: {{ tool_name }}."
+suite:
+  name: "suite_meryl"
+  description: "A suite of tools that brings the Meryl project into Galaxy."
+  long_description: |
+      Meryl a k-mer counter. It is built into the Celera assembler and is also available as a stand-alone application. Meryl uses a sorting-based approach that sorts k-mers in lexicographical order.
diff --git a/tools/meryl/arithmetic-kmers.xml b/tools/meryl/arithmetic-kmers.xml
index 565063019e9..b8eeabfd9ed 100644
--- a/tools/meryl/arithmetic-kmers.xml
+++ b/tools/meryl/arithmetic-kmers.xml
@@ -27,15 +27,14 @@
             <option value="decrease">Decrease: subtract x from the count of each k-mer</option>
             <option value="multiply">Multiply: multiply the count of each k-mer by x</option>
             <option value="divide">Divide: divide the count of each k-mer by x</option>
-            <option value="divide-round">Divide-round: divide the count of each k-mer by x and round th results</option>
+            <option value="divide-round">Divide-round: divide the count of each k-mer by x and round the results</option>
             <option value="modulo">Modulo: set the count of each k-mer to the remainder of the count divided by x</option>
         </param>
         <param name="input_meryldb_02" type="data" format="meryldb" label="Input meryldb" help="Select a meryldb dataset"/>
         <param name="X" type="integer" min="1" max="1000000" value="" optional="true" label="Operand"/>
     </inputs>
     <outputs>
-        <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb" label="${tool.name} on ${on_string}: read-db.meryldb">
-            </data>
+        <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb"/>
     </outputs>
     <tests>
         <test expect_num_outputs="1">
@@ -45,6 +44,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="59500" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -55,6 +55,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="42313" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -65,6 +66,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="60530" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -75,6 +77,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="56200" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -85,6 +88,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="56100" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -95,6 +99,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="37501" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
diff --git a/tools/meryl/count-kmers.xml b/tools/meryl/count-kmers.xml
index 4d24526e1ee..2f139da54a8 100644
--- a/tools/meryl/count-kmers.xml
+++ b/tools/meryl/count-kmers.xml
@@ -51,8 +51,7 @@
         </conditional>
     </inputs>
     <outputs>
-        <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb" label="${tool.name} on ${on_string}: read-db.meryldb">
-            </data>
+        <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb"/>
     </outputs>
     <tests>
         <test expect_num_outputs="1">
@@ -65,6 +64,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="22152" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
             <assert_stdout>
@@ -81,6 +81,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="22200" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
             <assert_stdout>
@@ -97,6 +98,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="22200" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
             <assert_stdout>
diff --git a/tools/meryl/filter-kmers.xml b/tools/meryl/filter-kmers.xml
index ecb7d12f4ef..275b71b8352 100644
--- a/tools/meryl/filter-kmers.xml
+++ b/tools/meryl/filter-kmers.xml
@@ -47,8 +47,7 @@
         </conditional>
     </inputs>
     <outputs>
-        <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb" label="${tool.name} on ${on_string}: read-db.meryldb">
-            </data>
+        <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb"/>
     </outputs>
     <tests>
         <test expect_num_outputs="1">
@@ -60,6 +59,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="32077" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -72,6 +72,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="49643" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -84,6 +85,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="1634" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -96,6 +98,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="2621" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -108,6 +111,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="59100" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
diff --git a/tools/meryl/groups-kmers.xml b/tools/meryl/groups-kmers.xml
index ca66eb112d8..d274262f557 100644
--- a/tools/meryl/groups-kmers.xml
+++ b/tools/meryl/groups-kmers.xml
@@ -36,8 +36,7 @@
         <param name="input_meryldb_02" type="data" multiple="true" format="meryldb" label="Input meryldb"/>
     </inputs>
     <outputs>
-        <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb" label="${tool.name} on ${on_string}: read-db.meryldb">
-            </data>
+        <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb"/>
     </outputs>
     <tests>
         <test expect_num_outputs="1">
@@ -46,6 +45,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="36100" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -55,6 +55,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="58925" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -64,6 +65,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="58930" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -73,6 +75,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="58600" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -82,6 +85,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="14951" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -91,6 +95,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="14957" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -100,6 +105,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="14956" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -109,6 +115,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="14953" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -118,6 +125,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="23999" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -127,6 +135,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="24016" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
@@ -136,6 +145,7 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="57455" delta="1000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
         </test>
diff --git a/tools/meryl/histogram-kmers.xml b/tools/meryl/histogram-kmers.xml
index a023b9bd7e7..0150263b381 100644
--- a/tools/meryl/histogram-kmers.xml
+++ b/tools/meryl/histogram-kmers.xml
@@ -20,13 +20,16 @@
         <param name="input_meryldb_02" type="data" format="meryldb" label="Input meryldb"/>
     </inputs>
     <outputs>
-        <data name="read_db_hist" format="tabular" from_work_dir="read-db.hist" label="${tool.name} on ${on_string}: read-db histogram">
-            </data>
+        <data name="read_db_hist" format="tabular" from_work_dir="read-db.hist"/>
     </outputs>
     <tests>
         <test expect_num_outputs="1">
             <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb"/>
-            <output name="read_db_hist" file="output_24.read-db.hist"/>
+            <output name="read_db_hist" file="output_24.read-db.hist">
+                <assert_contents>
+                    <expand macro="meryldb_archive_assumptions"/>
+                </assert_contents>
+            </output>
         </test>
     </tests>
     <help>
diff --git a/tools/meryl/macros.xml b/tools/meryl/macros.xml
index 85d23f65a6c..8c64a294031 100644
--- a/tools/meryl/macros.xml
+++ b/tools/meryl/macros.xml
@@ -17,6 +17,19 @@
             <requirement type="package" version="1.34">tar</requirement>
         </requirements>
     </xml>
+    <xml name="meryldb_archive_assumptions">
+        <has_archive_member path="^[^/]+.meryl$" n="1"/>  <!-- folder also counts -->
+        <has_archive_member path="[^/]+.meryl/merylIndex" n="1" all="true">
+            <has_size min="1"/>
+        </has_archive_member>
+        <has_archive_member path="[^/]+.meryl/.*\.merylIndex" n="64" all="true">
+            <has_size min="1"/>
+        </has_archive_member>
+        <has_archive_member path="[^/]+.meryl/.*\.merylData" n="64" all="true">
+            <!-- <has_size min="1"/> seems to be 0 sometimes -->
+        </has_archive_member>
+        <has_archive_member path=".*" n="130"/>
+    </xml>
     <xml name="citations">
         <citations>
             <citation type="bibtex">                                                                   
diff --git a/tools/meryl/print.xml b/tools/meryl/print.xml
index 66099114687..f2fe014a980 100644
--- a/tools/meryl/print.xml
+++ b/tools/meryl/print.xml
@@ -20,8 +20,7 @@
         <param name="input_meryldb_02" type="data" format="meryldb" label="Input meryldb"/>
     </inputs>
     <outputs>
-        <data name="read_db_print" format="tabular" from_work_dir="read-db.tabular" label="${tool.name} on ${on_string}: kmer counts">
-            </data>
+        <data name="read_db_print" format="tabular" from_work_dir="read-db.tabular"/>
     </outputs>
     <tests>
         <test expect_num_outputs="1">
diff --git a/tools/meryl/trio-mode.xml b/tools/meryl/trio-mode.xml
index 584f904389c..ba372acbc55 100644
--- a/tools/meryl/trio-mode.xml
+++ b/tools/meryl/trio-mode.xml
@@ -93,18 +93,12 @@
         </conditional>
     </inputs>
     <outputs>
-        <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb" label="${tool.name} on ${on_string}: read-db.meryldb">
-            </data>
-        <data name="read_db_hist" format="tabular" from_work_dir="read-db.hist" label="${tool.name} on ${on_string}: read-db histogram">
-            </data>
-        <data name="pat_db" format="meryldb" from_work_dir="pat.meryldb" label="${tool.name} on ${on_string}: pat.meryldb">
-            </data>
-        <data name="pat_db_hist" format="tabular" from_work_dir="pat.inherited.hist" label="${tool.name} on ${on_string}: paternal inherited histogram">
-            </data>
-        <data name="mat_db" format="meryldb" from_work_dir="mat.meryldb" label="${tool.name} on ${on_string}: mat.meryldb">
-            </data>
-        <data name="mat_db_hist" format="tabular" from_work_dir="mat.inherited.hist" label="${tool.name} on ${on_string}: maternal inherited histogram">
-            </data>
+        <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb" label="${tool.name} on ${on_string}: read-db.meryldb"/>
+        <data name="read_db_hist" format="tabular" from_work_dir="read-db.hist" label="${tool.name} on ${on_string}: read-db histogram"/>
+        <data name="pat_db" format="meryldb" from_work_dir="pat.meryldb" label="${tool.name} on ${on_string}: pat.meryldb"/>
+        <data name="pat_db_hist" format="tabular" from_work_dir="pat.inherited.hist" label="${tool.name} on ${on_string}: paternal inherited histogram"/>
+        <data name="mat_db" format="meryldb" from_work_dir="mat.meryldb" label="${tool.name} on ${on_string}: mat.meryldb"/>
+        <data name="mat_db_hist" format="tabular" from_work_dir="mat.inherited.hist" label="${tool.name} on ${on_string}: maternal inherited histogram"/>
     </outputs>
     <tests>
         <test expect_num_outputs="6">
@@ -118,18 +112,21 @@
             <output name="read_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="3362942" delta="2000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
             <output name="read_db_hist" file="output_23.read-db.hist"/>
             <output name="pat_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="120610" delta="400"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
             <output name="pat_db_hist" file="output_23.pat.hist"/>
             <output name="mat_db" ftype="meryldb">
                 <assert_contents>
                     <has_size value="67883" delta="300"/>
+                    <expand macro="meryldb_archive_assumptions"/>
                 </assert_contents>
             </output>
             <output name="mat_db_hist" file="output_23.mat.hist"/>