update genome download and labels for sourmash

fmalmeida · Jun 4, 2023 · 887f3a3 · 887f3a3
1 parent 88e0e06
commit 887f3a3
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 8 deletions.
diff --git a/modules/generic/ncbi_genome.nf b/modules/generic/ncbi_genome.nf
@@ -30,5 +30,11 @@ process GET_NCBI_GENOME {
         mv ncbi_dataset/data/*/*.fna . && \\
         rm -rf ncbi_dataset *.md *.txt
     done
+
+    # rename
+    for file in *.fna ; do
+        name=\$( echo \$file | cut -d '_' -f 1,2 ) ;
+        mv \$file \${name}.fna
+    done
     """
 }
diff --git a/modules/generic/sourmash_all.nf b/modules/generic/sourmash_all.nf
@@ -23,14 +23,17 @@ process SOURMASH_ALL {
 
     # sketch input genomes
     mkdir signatures ;
-    for genome in genomes/* ; do
-        export name=\$( echo \${genome} | cut -f 2 -d '/' | cut -f 1,2 -d '_' ) ; 
-        sourmash \\
-            sketch dna \\
-            -p scaled=${scale},k=${kmer} \\
-            \${genome} \\
-            -o signatures/\${name}.sig ;
-    done
+    ( 
+        cd genomes && \\
+        for genome in * ; do
+            export name=\$( echo \${genome} | cut -f 2 -d '/' | cut -f 1,2 -d '_' ) ; 
+            sourmash \\
+                sketch dna \\
+                -p scaled=${scale},k=${kmer} \\
+                \${genome} \\
+                -o ../signatures/\${name}.sig ;
+        done ;
+    )
 
     # compare
     sourmash \\
@@ -40,9 +43,11 @@ process SOURMASH_ALL {
         -o sourmash_cmp
     
     # plot
+    sourmash plot --labels sourmash_cmp
     sourmash \\
         plot \\
         --pdf \\
+        --csv sourmash_plot.csv \\
         --labels sourmash_cmp
     """
 }