Update_jabfeb2023

.gitignore

@@ -8,3 +8,4 @@ work
 trace*
 results/*
 Input_prot_gotits.txt
+blastdb

bin/blast2taxgenesummary.pl

@@ -8,7 +8,7 @@
 
 my $input = $ARGV[0];
 
-my $OUT_file="Horizontal_assesment.tsv";
+my $OUT_file="$input\_summary.tsv";
 
 #initiaite groups from conversion file.
 open(my $INPUT_IN, "<", $input)   or die "Could not open $input \n";

bin/tophitsonly.pl

@@ -0,0 +1,48 @@
+#!/usr/bin/perl
+#plot_event.pl
+use strict;
+use warnings;
+
+die "Please specify (1) Blast hits\n" unless(@ARGV==1);
+
+
+my $input = $ARGV[0];
+
+my $OUT_file="tophitsonly.tsv";
+
+#initiaite groups from conversion file.
+open(my $INPUT_IN, "<", $input)   or die "Could not open $input \n";
+open(my $outhandle, ">", $OUT_file)   or die "Could not open $OUT_file\n";
+
+my %genes;
+
+while (my $line=<$INPUT_IN>){
+	chomp $line;
+	my @split= split("\t", $line);
+
+	my $id= $split[0];
+	my $perc= $split[3];
+
+	if ($genes{$id}){
+		my $oldline=$genes{$id};
+		my @split2= split("\t", $oldline);
+
+		if ($perc > $split2[3]){
+			$genes{$id}="$line";
+		}
+		else{
+			#do nothing.
+		}
+
+	}
+	else{
+		$genes{$id}="$line";
+	}
+
+}
+
+foreach my $key ( sort keys %genes){
+
+	print $outhandle "$genes{$key}\n";
+
+}

conf/myriad.config

@@ -9,7 +9,7 @@ process {
     penv = "smp"
     cpus = 6
     container = 'chriswyatt/diamond'
-    clusterOptions = '-l mem=40G -l h_rt=24:0:0 -l tmpfs=400G'
+    clusterOptions = '-l mem=20G -l h_rt=48:0:0 -l tmpfs=200G'
   }
    withLabel: "blastdb" {
     penv = "smp"

main.nf

@@ -13,6 +13,8 @@ params.predownloaded= false
 params.outdir = "results"
 params.names = false
 params.nodes = false
+params.numhits = null
+params.tophits = null
 params.level = "family"
 params.sensitivity = "fast"
 params.horizontal = false
@@ -88,11 +90,12 @@ workflow {
 
 
 	DIAMOND_BLAST ( input_target_proteins , input_database )
+
 	PLOT_PIE ( input_nodes , input_names , DIAMOND_BLAST.out.blast_hits )
 
-	if ( params.horizontal ){
-		DIAMOND_HORIZONTAL ( input_target_proteins , input_database )
-	}
+	//if ( params.horizontal ){
+	//	DIAMOND_HORIZONTAL ( input_target_proteins , input_database )
+	//}
 
 	if ( params.xml ){
 		DIAMOND_XML ( input_target_proteins , input_database )

modules/diamond_blast.nf

@@ -12,8 +12,8 @@ process DIAMOND_BLAST {
 
     script:
     """
-    diamond blastp --$params.sensitivity --max-target-seqs $params.numhits --query $proteins --db nr --out ${proteins}\_results.tsv --threads $task.cpus --outfmt 6 qseqid sseqid stitle pident evalue sphylums staxids
-    #rm nr.dmnd
-    #rm $proteins
+	diamond blastp --$params.sensitivity --top $params.tophits --query $proteins --db nr --out ${proteins}\_results.tsv --threads $task.cpus --outfmt 6 qseqid sseqid stitle pident evalue sphylums staxids
+    	#rm nr.dmnd
+    	#rm $proteins
     """
 }

modules/plot_taxonomy_pie.nf

@@ -1,7 +1,8 @@
 process PLOT_PIE {
     label 'perl_pie'
-    publishDir "$params.outdir/Blast_results/"
-    //stageInMode 'copy'
+    publishDir "$params.outdir/Blast_results/", mode:'copy', pattern: '*top.tsv'
+    publishDir "$params.outdir/Taxo_figure/", mode:'copy', pattern: '*.pdf'
+    publishDir "$params.outdir/Taxo_summary/", mode:'copy', pattern: '*_summary.tsv'
 
     input:
 	path nodes
@@ -15,6 +16,10 @@ process PLOT_PIE {
 
     script:
     """
-	${workflow.projectDir}/bin/ncbi_txids_taxonomy.all.pl $nodes $names $blast_result
+	${workflow.projectDir}/bin/tophitsonly.pl $blast_result
+	mv tophitsonly.tsv ${blast_result}_top.tsv
+	${workflow.projectDir}/bin/ncbi_txids_taxonomy.all.pl $nodes $names ${blast_result}_top.tsv
+
+	blast2taxgenesummary.pl $blast_result
     """
 }

modules/transdecoder.nf

@@ -1,5 +1,6 @@
 process T_DECODER {
     label 'transdecoder'
+    publishDir "$params.outdir/Prot/", mode:'copy', pattern: '*.prot.fa'
 
     input:
 	path fasta_file