From 1353ce0afa474bf6b8f0372072ff624f882fba26 Mon Sep 17 00:00:00 2001
From: Erin Young <erin.olde@gmail.com>
Date: Wed, 12 Jul 2023 09:33:37 -0600
Subject: [PATCH] added more metrics

---
 modules/snp-dists.nf | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/modules/snp-dists.nf b/modules/snp-dists.nf
index 186e663..3728298 100644
--- a/modules/snp-dists.nf
+++ b/modules/snp-dists.nf
@@ -10,10 +10,12 @@ process snp_dists {
   //#UPHLICA time '24h'
   
   input:
-  file(contigs)
+  tuple file(contigs), val(num_samples), val(num_core_genes)
 
   output:
   path "snp-dists/snp_matrix.txt"                                      , emit: snp_matrix
+  path "snp-dists/snp_matrix_with_qc.txt"
+  path "snp-dists/roary_metrics_mqc.csv"                               , emit: for_multiqc                                      
   path "logs/${task.process}/${task.process}.${workflow.sessionId}.log", emit: log_files
 
   shell:
@@ -31,5 +33,12 @@ process snp_dists {
     snp-dists !{params.snp_dists_options} \
       !{contigs} \
       > snp-dists/snp_matrix.txt
+
+    genome_length=$(cat !{contigs} | tr "\n" ";" | sed 's/>[^>]*//2g' | tr ";" "\n" | grep -v ">" | wc -c )
+
+    sed '0,/,/s/,/num_samples=!{num_samples};num_core_genes=!{num_core_genes};core_genome_length=$genome_length,/' snp-dists/snp_matrix.txt > snp-dists/snp_matrix_with_qc.txt
+
+    echo "num_samples,num_core_genes,core_genome_length"    >  snp-dists/roary_metrics_mqc.csv
+    echo "!{num_samples},!{num_core_genes},$genome_length" >> snp-dists/roary_metrics_mqc.csv
   '''
 }