From fb0e4bf79f1ed3e74f5677d518f99b399dd91698 Mon Sep 17 00:00:00 2001
From: Peter Kruczkiewicz <peter.kruczkiewicz@gmail.com>
Date: Thu, 25 Jul 2024 09:32:45 -0500
Subject: [PATCH] feat: create files summarizing annotation issues and failures

---
 conf/modules_illumina.config | 11 +++++++++++
 conf/modules_nanopore.config | 11 +++++++++++
 modules/local/vadr.nf        | 18 ++++++++++++++++++
 workflows/illumina.nf        |  3 ++-
 workflows/nanopore.nf        |  3 ++-
 5 files changed, 44 insertions(+), 2 deletions(-)

diff --git a/conf/modules_illumina.config b/conf/modules_illumina.config
index 4aa6041..eee9939 100644
--- a/conf/modules_illumina.config
+++ b/conf/modules_illumina.config
@@ -37,6 +37,17 @@ process {
     ]
   }
 
+  withName: 'VADR_SUMMARIZE_ISSUES' {
+    ext.args = ''
+    publishDir = [
+      [
+        path: { "${params.outdir}/annotation" },
+        mode: params.publish_dir_mode,
+        saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+      ]
+    ]
+  }
+
   withName: 'POST_TABLE2ASN' {
     ext.args = ''
     publishDir = [
diff --git a/conf/modules_nanopore.config b/conf/modules_nanopore.config
index eba39e9..712e756 100644
--- a/conf/modules_nanopore.config
+++ b/conf/modules_nanopore.config
@@ -255,6 +255,17 @@ process {
     ]
   }
 
+  withName: 'VADR_SUMMARIZE_ISSUES' {
+    ext.args = ''
+    publishDir = [
+      [
+        path: { "${params.outdir}/annotation" },
+        mode: params.publish_dir_mode,
+        saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+      ]
+    ]
+  }
+
   withName: 'POST_TABLE2ASN' {
     ext.args = ''
     publishDir = [
diff --git a/modules/local/vadr.nf b/modules/local/vadr.nf
index c11e93e..933a4ea 100644
--- a/modules/local/vadr.nf
+++ b/modules/local/vadr.nf
@@ -29,3 +29,21 @@ process VADR {
   END_VERSIONS
   """
 }
+
+process VADR_SUMMARIZE_ISSUES {
+  executor 'local'
+  memory 100.MB
+
+  input:
+  path(vadr_output, stageAs: "input*/*")
+
+  output:
+  path('vadr-annotation-issues.txt'), emit: issues
+  path('vadr-annotation-failed-sequences.txt'), emit: failed
+
+  script:
+  """
+  cat input*/**/*.alt.list | awk 'NR == 1 || \$0 !~ /^#/' > vadr-annotation-issues.txt
+  cat input*/**/*.fail.list > vadr-annotation-failed-sequences.txt
+  """
+}
diff --git a/workflows/illumina.nf b/workflows/illumina.nf
index 8c92b2a..5137593 100644
--- a/workflows/illumina.nf
+++ b/workflows/illumina.nf
@@ -20,7 +20,7 @@ include { BLAST_MAKEBLASTDB } from '../modules/local/blast_makeblastdb'
 include { BLAST_BLASTN } from '../modules/local/blastn'
 include { CAT_ILLUMINA_FASTQ } from '../modules/local/cat_illumina_fastq'
 include { ZSTD_DECOMPRESS as ZSTD_DECOMPRESS_FASTA; ZSTD_DECOMPRESS as ZSTD_DECOMPRESS_CSV } from '../modules/local/zstd_decompress'
-include { VADR                                                } from '../modules/local/vadr'
+include { VADR; VADR_SUMMARIZE_ISSUES                         } from '../modules/local/vadr'
 include { PRE_TABLE2ASN; TABLE2ASN; POST_TABLE2ASN            } from '../modules/local/table2asn'
 include { CUSTOM_DUMPSOFTWAREVERSIONS  as SOFTWARE_VERSIONS   } from '../modules/nf-core/modules/custom/dumpsoftwareversions/main'
 
@@ -92,6 +92,7 @@ workflow ILLUMINA {
   VADR.out.feature_table
     .combine(VADR.out.pass_fasta, by: 0)
     .set { ch_pre_table2asn }
+  VADR_SUMMARIZE_ISSUES(VADR.out.vadr_outdir.map { [it[1]] }.collect())
   PRE_TABLE2ASN(ch_pre_table2asn)
   ch_versions = ch_versions.mix(PRE_TABLE2ASN.out.versions)
   TABLE2ASN(PRE_TABLE2ASN.out.table2asn_input)
diff --git a/workflows/nanopore.nf b/workflows/nanopore.nf
index 91641ae..96a82bc 100644
--- a/workflows/nanopore.nf
+++ b/workflows/nanopore.nf
@@ -32,7 +32,7 @@ include { BLAST_BLASTN as BLAST_BLASTN_IRMA                   } from '../modules
 include { BLAST_BLASTN as BLAST_BLASTN_CONSENSUS              } from '../modules/local/blastn'
 include { BLAST_BLASTN as BLAST_BLASTN_CONSENSUS_REF_DB       } from '../modules/local/blastn'
 include { CUSTOM_DUMPSOFTWAREVERSIONS  as SOFTWARE_VERSIONS   } from '../modules/nf-core/modules/custom/dumpsoftwareversions/main'
-include { VADR                                                } from '../modules/local/vadr'
+include { VADR; VADR_SUMMARIZE_ISSUES                         } from '../modules/local/vadr'
 include { PRE_TABLE2ASN; TABLE2ASN; POST_TABLE2ASN            } from '../modules/local/table2asn'
 include { MULTIQC                                             } from '../modules/local/multiqc'
 
@@ -247,6 +247,7 @@ workflow NANOPORE {
   VADR.out.feature_table
     .combine(VADR.out.pass_fasta, by: 0)
     .set { ch_pre_table2asn }
+  VADR_SUMMARIZE_ISSUES(VADR.out.vadr_outdir.map { [it[1]] }.collect())
   PRE_TABLE2ASN(ch_pre_table2asn)
   ch_versions = ch_versions.mix(PRE_TABLE2ASN.out.versions)
   TABLE2ASN(PRE_TABLE2ASN.out.table2asn_input)