Merge pull request #190 from CDCgov/user_provided_annotations_kao

User provided annotations for genbank submission
CDCgov · Apr 4, 2024 · 660bdb7 · 660bdb7
2 parents 8e7cf04 + b462cfc
commit 660bdb7
Show file tree

Hide file tree

Showing 14 changed files with 1,177 additions and 35 deletions.
diff --git a/...fs/mpox_split_gffs/OH0002_reformatted.gff → ...ple_annotations/Cdiphtheriae/BX248355.gff b/...fs/mpox_split_gffs/OH0002_reformatted.gff → ...ple_annotations/Cdiphtheriae/BX248355.gff
diff --git a/...fs/mpox_split_gffs/TX0001_reformatted.gff → ...ple_annotations/Cdiphtheriae/CP014051.gff b/...fs/mpox_split_gffs/TX0001_reformatted.gff → ...ple_annotations/Cdiphtheriae/CP014051.gff
diff --git a/...fs/mpox_split_gffs/NY0007_reformatted.gff → ...ple_annotations/Cdiphtheriae/CP040557.gff b/...fs/mpox_split_gffs/NY0007_reformatted.gff → ...ple_annotations/Cdiphtheriae/CP040557.gff
diff --git a/...fs/mpox_split_gffs/FL0004_reformatted.gff → ...e_annotations/mpox/FL0004_reformatted.gff b/...fs/mpox_split_gffs/FL0004_reformatted.gff → ...e_annotations/mpox/FL0004_reformatted.gff
diff --git a/...fs/mpox_split_gffs/FL0015_reformatted.gff → ...e_annotations/mpox/FL0015_reformatted.gff b/...fs/mpox_split_gffs/FL0015_reformatted.gff → ...e_annotations/mpox/FL0015_reformatted.gff
diff --git a/...fs/mpox_split_gffs/IL0005_reformatted.gff → ...e_annotations/mpox/IL0005_reformatted.gff b/...fs/mpox_split_gffs/IL0005_reformatted.gff → ...e_annotations/mpox/IL0005_reformatted.gff
diff --git a/...fs/mpox_split_gffs/NY0006_reformatted.gff → ...e_annotations/mpox/NY0006_reformatted.gff b/...fs/mpox_split_gffs/NY0006_reformatted.gff → ...e_annotations/mpox/NY0006_reformatted.gff
diff --git a/assets/sample_annotations/mpox/NY0007_reformatted.gff b/assets/sample_annotations/mpox/NY0007_reformatted.gff
diff --git a/assets/sample_annotations/mpox/OH0002_reformatted.gff b/assets/sample_annotations/mpox/OH0002_reformatted.gff
diff --git a/assets/sample_annotations/mpox/TX0001_reformatted.gff b/assets/sample_annotations/mpox/TX0001_reformatted.gff
diff --git a/...split_gffs/VARV_RZ10_3587_reformatted.gff → ...ns/variola/VARV_RZ10_3587_reformatted.gff b/...split_gffs/VARV_RZ10_3587_reformatted.gff → ...ns/variola/VARV_RZ10_3587_reformatted.gff
diff --git a/assets/sample_metadata/MPXV_user_provided_annotations.xlsx b/assets/sample_metadata/MPXV_user_provided_annotations.xlsx
diff --git a/assets/sample_metadata/custom_fields_MPXV_metadata_Sample_Run_1.xlsx b/assets/sample_metadata/custom_fields_MPXV_metadata_Sample_Run_1.xlsx
diff --git a/workflows/tostadas.nf b/workflows/tostadas.nf
@@ -74,12 +74,24 @@ workflow TOSTADAS {
             fastq1 = row.fastq_path_1 ? file(row.fastq_path_1) : null
             fastq2 = row.fastq_path_2 ? file(row.fastq_path_2) : null
             meta = [id:row.sequence_name]
-            [meta, fasta_path, fastq1, fastq2]
+            gff = row.gff_path ? file(row.gff_path) : null
+            if (gff == null) {
+                [meta, fasta_path, fastq1, fastq2]
+            }
+            else {
+                [meta, fasta_path, fastq1, fastq2, gff]
+            }
         }
 
     // Create initial submission channel
     submission_ch = metadata_ch.join(reads_ch)
 
+    // print error if they provide gff and annotation flag
+    // if ( !params.annotation && params.genbank ) {
+    //             // todo: make an error msg that follows the rest of the code protocol
+    //             throw new Exception("Cannot submit to GenBank without assembly and annotation files")
+    //     }
+
     // check if the user wants to skip annotation or not
     if ( params.annotation ) {
         if ( params.virus && !params.bacteria ) {
@@ -137,46 +149,18 @@ workflow TOSTADAS {
 
     // run submission for the annotated samples 
     if ( params.submission ) {
-
         // pre submission process + get wait time (parallel)
         GET_WAIT_TIME (
             METADATA_VALIDATION.out.tsv_Files.collect() 
         )
-
-        // check if annotation is set to true 
-        if ( params.annotation ) {   
-            if (params.sra && params.genbank ) {                     // sra and genbank
-                INITIAL_SUBMISSION (
-                    submission_ch,  // meta.id, fasta, fastq1, fastq2, gff
-                    params.submission_config,  
-                    GET_WAIT_TIME.out
-                    )
-                } 
-            else {      
-                if (! params.sra && params.genbank ) {               // only genebankk
-                    INITIAL_SUBMISSION ( 
-                        submission_ch,     // meta.id, fasta, "", "", gff
-                        params.submission_config, 
-                        GET_WAIT_TIME.out
-                        )
-                    }
-                }
-        }
-        }
-        if ( !params.annotation && params.sra ) {        // no annotation only fastq submission
-
-            INITIAL_SUBMISSION (
-                submission_ch,       // meta.id, "", fastq1, fastq2, gff
-                params.submission_config, 
-                GET_WAIT_TIME.out
+        INITIAL_SUBMISSION (
+            submission_ch,  // meta.id, fasta, fastq1, fastq2, gff
+            params.submission_config,  
+            GET_WAIT_TIME.out
             )
-        } 
-        if ( !params.annotation && params.genbank ) {
-                // todo: make an error msg that follows the rest of the code protocol
-                throw new Exception("Cannot submit to GenBank without assembly and annotation files")
-        }
 
-        // todo test update submission
+        }
+        // to do remove if not needed
         if ( params.update_submission ) {
             UPDATE_SUBMISSION (
                 '',