Adding in flag for no wobble / just exact duplicates (92de655d) · Commits · Chaos / chaos.juicer

SLURM/scripts/split_rmdups.awk

+16 −6

Changes for SLURM/scripts/split_rmdups.awk: 16 added lines, 6 removed lines.
Original line number	Diff line number	Diff line
		@@ -35,7 +35,12 @@ BEGIN{
		if (p1 != $1 \|\| p2 != $2 \|\| p4 != $4 \|\| p5 != $5 \|\| p8 != $8) {
		sname = sprintf("%s_msplit%04d_", groupname, name);
		sscriptname = sprintf("%s/.%s.slurm", debugdir, sname);
		if (justexact) {
		printf("#!/bin/bash -l\n#SBATCH -o %s/dup-split-%s.out\n#SBATCH -e %s/dup-split-%s.err\n#SBATCH -p %s\n#SBATCH -J %s_msplit0\n#SBATCH -t 1440\n#SBATCH -c 1\n#SBATCH --ntasks=1\ndate;awk -f %s/scripts/dups.awk -v nowobble=1 -v name=%s/%s %s/split%04d;\necho Reads:%s\ndate\n", debugdir, name, debugdir, name, queue, groupname, juicedir, dir, sname, dir, name, tot) > sscriptname;
		}
		else {
		printf("#!/bin/bash -l\n#SBATCH -o %s/dup-split-%s.out\n#SBATCH -e %s/dup-split-%s.err\n#SBATCH -p %s\n#SBATCH -J %s_msplit0\n#SBATCH -t 1440\n#SBATCH -c 1\n#SBATCH --ntasks=1\ndate;awk -f %s/scripts/dups.awk -v name=%s/%s %s/split%04d;\necho Reads:%s\ndate\n", debugdir, name, debugdir, name, queue, groupname, juicedir, dir, sname, dir, name, tot) > sscriptname;
		}
		sysstring = sprintf("sbatch %s", sscriptname);
		system(sysstring);
		outname = sprintf("%s/split%04d", dir, name);
		@@ -54,22 +59,27 @@ BEGIN{
		END {
		sname = sprintf("%s_msplit%04d_", groupname, name);
		sscriptname = sprintf("%s/.%s.slurm", debugdir, sname);
		if (justexact) {
		printf("#!/bin/bash -l\n#SBATCH -o %s/dup-split-%s.out\n#SBATCH -e %s/dup-split-%s.err\n#SBATCH -p %s\n#SBATCH -J %s_msplit0\n#SBATCH -t 1440\n#SBATCH -c 1\n#SBATCH --ntasks=1\ndate;awk -f %s/scripts/dups.awk -v nowobble=1 -v name=%s/%s %s/split%04d;\necho Reads:%s\ndate\n", debugdir, name, debugdir, name, queue, groupname, juicedir, dir, sname, dir, name, tot) > sscriptname;
		}
		else {
		printf("#!/bin/bash -l\n#SBATCH -o %s/dup-split-%s.out\n#SBATCH -e %s/dup-split-%s.err\n#SBATCH -p %s\n#SBATCH -J %s_msplit0\n#SBATCH -t 1440\n#SBATCH -c 1\n#SBATCH --ntasks=1\ndate;awk -f %s/scripts/dups.awk -v name=%s/%s %s/split%04d;\necho Reads:%s\ndate\n", debugdir, name, debugdir, name, queue, groupname, juicedir, dir, sname, dir, name, tot) > sscriptname;
		}
		sysstring = sprintf("sbatch %s", sscriptname);
		system(sysstring);
		close(sscriptname);

		sscriptname = sprintf("%s/.%s_msplit.slurm", debugdir, groupname);
		printf("#!/bin/bash -l\n#SBATCH -o %s/dup-merge.out\n#SBATCH -e %s/dup-merge.err\n#SBATCH -p %s\n#SBATCH -J %s_msplit0\n#SBATCH -d singleton\n#SBATCH -t 1440\n#SBATCH -c 1\n#SBATCH --ntasks=1\ndate;\necho \"\"; cat %s/%s_msplit_optdups.txt > %s/opt_dups.txt;cat %s/%s_msplit_dups.txt > %s/dups.txt;cat %s/%s_msplit*_merged_nodups.txt > %s/merged_nodups.txt;\ndate\n", debugdir, debugdir, queue, groupname, dir, groupname, dir, dir, groupname, dir, dir, groupname, dir) > sscriptname;
		printf("#!/bin/bash -l\n#SBATCH -o %s/dup-merge.out\n#SBATCH -e %s/dup-merge.err\n#SBATCH --mem=50G\n#SBATCH -p %s\n#SBATCH -J %s_msplit0\n#SBATCH -d singleton\n#SBATCH -t 1440\n#SBATCH -c 1\n#SBATCH --ntasks=1\ndate;\necho \"\"; cat %s/%s_msplit_optdups.txt > %s/opt_dups.txt;cat %s/%s_msplit_dups.txt > %s/dups.txt;cat %s/%s_msplit*_merged_nodups.txt > %s/merged_nodups.txt;\ndate\n", debugdir, debugdir, queue, groupname, dir, groupname, dir, dir, groupname, dir, dir, groupname, dir) > sscriptname;
		sysstring = sprintf("sbatch %s", sscriptname);
		system(sysstring);
		close(sscriptname);

		sscriptname = sprintf("%s/.%s_rmsplit.slurm", debugdir, groupname);
		printf("#!/bin/bash -l\n#SBATCH -o %s/dup-rm.out\n#SBATCH -e %s/dup-rm.err\n#SBATCH -p %s\n#SBATCH -J %s_msplit0\n#SBATCH -d singleton\n#SBATCH -t 1440\n#SBATCH -c 1\n#SBATCH --ntasks=1\ndate;\nrm %s/_msplit_optdups.txt; rm %s/_msplit_dups.txt; rm %s/_msplit_merged_nodups.txt;rm %s/split*;\ndate\n", debugdir, debugdir, queue, groupname, dir, dir, dir, dir) > sscriptname;
		sysstring = sprintf("sbatch %s", sscriptname);
		system(sysstring);
		close(sscriptname);
		# sscriptname = sprintf("%s/.%s_rmsplit.slurm", debugdir, groupname);
		# printf("#!/bin/bash -l\n#SBATCH -o %s/dup-rm.out\n#SBATCH -e %s/dup-rm.err\n#SBATCH -p %s\n#SBATCH -J %s_msplit0\n#SBATCH -d singleton\n#SBATCH -t 1440\n#SBATCH -c 1\n#SBATCH --ntasks=1\ndate;\nrm %s/_msplit_optdups.txt; rm %s/_msplit_dups.txt; rm %s/_msplit_merged_nodups.txt;rm %s/split*;\ndate\n", debugdir, debugdir, queue, groupname, dir, dir, dir, dir) > sscriptname;
		# sysstring = sprintf("sbatch %s", sscriptname);
		# system(sysstring);
		# close(sscriptname);

		sscriptname = sprintf("%s/.%s_finalize.slurm", debugdir, groupname);
		printf("#!/bin/bash -l\n#SBATCH -o %s/dup-guard-trigger.out\n#SBATCH -e %s/dup-guard-trigger.err\n#SBATCH -p %s\n#SBATCH -J %s_msplit0\n#SBATCH -d singleton\n#SBATCH -t 1440\n#SBATCH -c 1\n#SBATCH --ntasks=1\necho %s %s %s %s;\nsqueue -u %s;\ndate\n", debugdir, debugdir, queue, groupname, topDir, site, genomeID, genomePath, user, user) > sscriptname;

Admin message