add pept2taxa

lib/commands/unipept.ts

@@ -5,6 +5,7 @@ import { Pept2go } from './unipept/pept2go.js';
 import { Pept2interpro } from './unipept/pept2interpro.js';
 import { Pept2lca } from './unipept/pept2lca.js';
 import { Pept2prot } from './unipept/pept2prot.js';
+import { Pept2taxa } from './unipept/pept2taxa.js';
 
 export class Unipept extends BaseCommand {
 
@@ -29,7 +30,8 @@ The command will give priority to the first way the input is passed, in the orde
       .addCommand(new Pept2go().command)
       .addCommand(new Pept2interpro().command)
       .addCommand(new Pept2lca().command)
-      .addCommand(new Pept2prot().command);
+      .addCommand(new Pept2prot().command)
+      .addCommand(new Pept2taxa().command);
   }
 
   async run(args?: string[]) {

lib/commands/unipept/pept2interpro.ts

@@ -15,7 +15,7 @@ The command will give priority to the first way tryptic peptides are passed, in
     super("pept2interpro");
 
     this.command
-      .summary("Fetch GO terms of UniProt entries that match tryptic peptides.")
+      .summary("Fetch InterPro entries of UniProt entries that match tryptic peptides.")
       .description(this.description)
       .option("-e, --equate", "equate isoleucine (I) and leucine (L) when matching peptides")
       .option("-a, --all", "Also return the names of the InterPro entries. Note that this may have a performance penalty.")

lib/commands/unipept/pept2prot.ts

@@ -18,7 +18,7 @@ The command will give priority to the first way tryptic peptides are passed, in
       .summary("Fetch UniProt entries that match tryptic peptides.")
       .description(this.description)
       .option("-e, --equate", "equate isoleucine (I) and leucine (L) when matching peptides")
-      .option("-a, --all", "Also return the names of the EC numbers. Note that this may have a performance penalty.")
+      .option("-a, --all", "report all information fields of UniProt entries available in Unipept. Note that this may have a performance penalty.")
       .addOption(new Option("-s --select <fields...>", "select the information fields to return. Selected fields are passed as a comma separated list of field names. Multiple -s (or --select) options may be used."))
       .argument("[peptides...]", "optionally, 1 or more peptides")
       .action((args, options) => this.run(args, options));

lib/commands/unipept/pept2taxa.ts

@@ -0,0 +1,34 @@
+import { Option } from "commander";
+import { UnipeptSubcommand } from "./unipept_subcommand.js";
+
+export class Pept2taxa extends UnipeptSubcommand {
+
+  readonly description = `For each tryptic peptide the unipept pept2taxa command retrieves from Unipept the set of taxa from all UniProt entries whose protein sequence contains an exact matches to the tryptic peptide. The command expects a list of tryptic peptides that are passed
+
+- as separate command line arguments
+- in a text file that is passed as an argument to the -i option
+- to standard input
+
+The command will give priority to the first way tryptic peptides are passed, in the order as listed above. Text files and standard input should have one tryptic peptide per line.`;
+
+  constructor() {
+    super("pept2taxa");
+
+    this.command
+      .summary("Fetch taxa of UniProt entries that match tryptic peptides.")
+      .description(this.description)
+      .option("-e, --equate", "equate isoleucine (I) and leucine (L) when matching peptides")
+      .option("-a, --all", "report all information fields of NCBI Taxonomy records available in Unipept. Note that this may have a performance penalty.")
+      .addOption(new Option("-s --select <fields...>", "select the information fields to return. Selected fields are passed as a comma separated list of field names. Multiple -s (or --select) options may be used."))
+      .argument("[peptides...]", "optionally, 1 or more peptides")
+      .action((args, options) => this.run(args, options));
+  }
+
+  requiredFields(): string[] {
+    return ["peptide"];
+  }
+
+  defaultBatchSize(): number {
+    return 5;
+  }
+}

lib/commands/unipept/unipept_subcommand.ts

@@ -60,6 +60,12 @@ export abstract class UnipeptSubcommand {
     this.formatter = FormatterFactory.getFormatter(this.options.format);
     if (this.options.output) {
       this.outputStream = createWriteStream(this.options.output);
+    } else {
+      process.stdout.on("error", (err) => {
+        if (err.code === "EPIPE") {
+          process.exit(0);
+        }
+      })
     }
 
     const iterator = this.getInputIterator(args, options.input as string);

tests/commands/unipept/pept2taxa.test.ts

@@ -0,0 +1,27 @@
+import { jest } from '@jest/globals';
+import { Pept2taxa } from "../../../lib/commands/unipept/pept2taxa";
+
+let output: string[];
+jest
+  .spyOn(process.stdout, "write")
+  .mockImplementation((data: unknown) => { output.push(data as string); return true; });
+
+beforeEach(() => {
+  output = [];
+});
+
+test('test with default args', async () => {
+  const command = new Pept2taxa();
+  await command.run(["AALTER"], { header: true, format: "csv" });
+  expect(output[0].startsWith("peptide,taxon_id,taxon_name,taxon_rank")).toBeTruthy();
+  expect(output[1].startsWith("AALTER,41,Stigmatella aurantiaca,species")).toBeTruthy();
+  expect(output.length).toBe(2);
+});
+
+test('test with fasta', async () => {
+  const command = new Pept2taxa();
+  await command.run([">test", "AALTER"], { header: true, format: "csv" });
+  expect(output[0].startsWith("fasta_header,peptide,taxon_id,taxon_name,taxon_rank")).toBeTruthy();
+  expect(output[1].startsWith(">test,AALTER,41,Stigmatella aurantiaca,species")).toBeTruthy();
+  expect(output.length).toBe(2);
+});