Merge branch 'rf-roothog-inference' into dev

Author: alpae

.github/workflows/docker-image.yml

@@ -65,11 +65,21 @@ jobs:
           username: ${{ secrets.DOCKER_HUB_USERNAME }}
           password: ${{ secrets.DOCKER_HUB_ACCESS_TOKEN }}
 
+      - name: Set platforms
+        id: set_platforms
+        run: |
+          echo "github ref: ${GITHUB_REF}"
+          if [[ "${GITHUB_REF##*/}" == "main" || "${GITHUB_REF##*/}" == "dev" ]]; then
+            echo "platforms=linux/amd64,linux/arm64" >> $GITHUB_OUTPUT
+          else
+            echo "platforms=linux/amd64" >> $GITHUB_OUTPUT
+          fi
+
       - name: Build and push
         uses: docker/build-push-action@v6
         with:
           context: .
-          platforms: linux/amd64,linux/arm64
+          platforms: ${{ steps.set_platforms.outputs.platforms }}
           push: true
           #${{ github.event_name != 'push' && github.event_name != 'pull_request' }}
           tags: ${{ steps.meta.outputs.tags }}

FastOMA/_utils_roothog.py

@@ -202,14 +202,11 @@ def check_splice(isoform_by_gene_all):
 
     logger.debug("For "+str(spliced_species_num)+"  species, out of "+str(len(isoform_by_gene_all))+" , we have splice files.")
     if total_genes ==0:
-        logger.debug("No splice information found. It could be that each line of  splice files has only one item. Make sure that the splitter in each line is semicolon ; . Or the splice folder is empty. You need to delete the empty splice folder. FastOMA check input failed! Exiting...")
+        logger.error("No splice information found. It could be that each line of  splice files has only one item. Make sure that the splitter in each line is semicolon ; . Or the splice folder is empty. You need to delete the empty splice folder. FastOMA check input failed! Exiting...")
         sys.exit(1)
 
     logger.debug("In total, for "+ str(total_genes)+" genes, we have " + str(total_isoforms)+"  splices.")
 
-    # make sys back
-    return 1
-
 
 def fastoma_check_input():
     import argparse
@@ -275,7 +272,7 @@ def fastoma_check_input():
     splice_files = conf.splice is not None and os.path.exists(conf.splice)
     if splice_files:
         logger.debug("splice folder exist. Let's see its inside.")
-        isoform_by_gene_all = _utils_roothog.parse_isoform_file(species_names, conf.splice)
+        isoform_by_gene_all = _utils_roothog.parse_isoform_files(species_names, conf.splice)
         check_splice(isoform_by_gene_all)
     else:
         logger.info("Splice folder doesn't exist and that's ok.")

FastOMA/check_input.py

@@ -178,9 +178,11 @@ def write_hog_orthoxml(pickle_folder, output_xml_name, gene_id_pickle_file, id_t
         species_xml = ET.SubElement(orthoxml_file, "species", attrib={"name": query_species_name, "taxonId": str(name2taxid[query_species_name]), "NCBITaxId": "0"})
         database_xml = ET.SubElement(species_xml, "database", attrib={"name": "database", "version": "2023"})
         genes_xml = ET.SubElement(database_xml, "genes")
-        for (gene_idx_integer, query_prot_name) in list_prots:
-            prot_id = id_transformer.transform(query_prot_name)
-            gene_xml = ET.SubElement(genes_xml, "gene", attrib={"id": str(gene_idx_integer), "protId": prot_id})
+        for gene in list_prots:
+            attribs = {"id": str(gene.numeric_id), "protId": id_transformer.transform(gene.prot_id)}
+            if gene.main_isoform is not None:
+                attribs["main_isoform"] = str(gene.main_isoform)
+            gene_xml = ET.SubElement(genes_xml, "gene", attrib=attribs)
     logger.debug("gene_xml is created.")
     orthoxml_file.append(taxonomy)