annotate scripts/stx_subtype_pe.sh @ 6:20ff3dca457f draft default tip

planemo upload commit 6857c749c21f580c828aba3543e294b69d32b662
author iss
date Mon, 23 Oct 2023 11:45:36 +0000
parents c6bab5103a14
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
1 tooldir="$1";
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
2 fastqfile1="$2";
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
3 fastqfile2="$3";
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
4 fastafile="$4";
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
5
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
6 # ASSEMBLY
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
7 mkdir stxdir;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
8 skesa --fastq $fastqfile1 $fastqfile2 --contigs_out stxdir/skesa.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
9 cp $fastafile stxdir/spades.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
10 rm -r output_dir;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
11
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
12 # FILTER + ASSEMBLY
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
13 chmod u+x $tooldir/scripts/duk
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
14 $tooldir/scripts/duk -m stxdir/filtered1STX.fq -k 23 $tooldir/data/stx.fa $fastqfile1;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
15 $tooldir/scripts/duk -m stxdir/filtered2STX.fq -k 23 $tooldir/data/stx.fa $fastqfile2;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
16 $tooldir/scripts/fastq_pair stxdir/filtered1STX.fq stxdir/filtered2STX.fq;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
17 $tooldir/scripts/fastq_pair stxdir/filtered1STX.fq.single.fq $fastqfile2;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
18 $tooldir/scripts/fastq_pair stxdir/filtered2STX.fq.single.fq $fastqfile1;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
19 cat stxdir/filtered1STX.fq.paired.fq > stxdir/filtered1STX_paired.fq;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
20 cat stxdir/filtered1STX.fq.single.fq.paired.fq >> stxdir/filtered1STX_paired.fq;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
21 cat $fastqfile1.paired.fq >> stxdir/filtered1STX_paired.fq;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
22 cat stxdir/filtered2STX.fq.paired.fq > stxdir/filtered2STX_paired.fq;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
23 cat $fastqfile2.paired.fq >> stxdir/filtered2STX_paired.fq;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
24 cat stxdir/filtered2STX.fq.single.fq.paired.fq >> stxdir/filtered2STX_paired.fq;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
25 dukstx1filesize=$(wc -c "stxdir/filtered1STX_paired.fq" | awk '{print $1}');
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
26 dukstx2filesize=$(wc -c "stxdir/filtered2STX_paired.fq" | awk '{print $1}');
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
27 if [ $dukstx1filesize -gt 0 ] && [ $dukstx2filesize -gt 0 ]
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
28 then
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
29 skesa --fastq stxdir/filtered1STX_paired.fq stxdir/filtered2STX_paired.fq --contigs_out stxdir/duk_skesa.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
30 perl $tooldir/scripts/spades.pl duk_spades_contigs duk_spades_contig_stats duk_spades_scaffolds duk_spades_scaffold_stats duk_spades_log NODE spades.py --disable-gzip-output --isolate -t 8 --pe1-ff --pe1-1 fastq:stxdir/filtered1STX_paired.fq --pe1-2 fastq:stxdir/filtered2STX_paired.fq
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
31 mv duk_spades_contigs stxdir/duk_spades.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
32 rm -r output_dir;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
33 blastn -query stxdir/duk_skesa.fasta -db $tooldir/data/stx -task blastn -evalue 0.001 -out stxdir/duk_skesa_seqs -outfmt '6 qseqid sseqid sframe qseq' -num_threads 8 -strand both -dust yes -max_target_seqs 1 -perc_identity 95.0;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
34 blastn -query stxdir/duk_spades.fasta -db $tooldir/data/stx -task blastn -evalue 0.001 -out stxdir/duk_spades_seqs -outfmt '6 qseqid sseqid sframe qseq' -num_threads 8 -strand both -dust yes -max_target_seqs 1 -perc_identity 95.0;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
35 else
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
36 touch stxdir/duk_skesa_seqs;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
37 touch stxdir/duk_spades_seqs;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
38 fi
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
39
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
40 # SEQUENCE SEARCH
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
41 blastn -query stxdir/skesa.fasta -db $tooldir/data/stx -task blastn -evalue 0.001 -out stxdir/skesa_seqs -outfmt '6 qseqid sseqid sframe qseq' -num_threads 8 -strand both -dust yes -max_target_seqs 1 -perc_identity 95.0;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
42 blastn -query stxdir/spades.fasta -db $tooldir/data/stx -task blastn -evalue 0.001 -out stxdir/spades_seqs -outfmt '6 qseqid sseqid sframe qseq' -num_threads 8 -strand both -dust yes -max_target_seqs 1 -perc_identity 95.0;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
43 # DIVIDE STX1 FROM STX2
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
44 awk 'BEGIN { OFS="\t" }!seen[$1]++ {if (substr($2,1,4)=="stx1") print $1,$3,$4}' stxdir/skesa_seqs > stxdir/stx1_skesa_seqs;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
45 awk 'BEGIN { OFS="\t" }!seen[$1]++ {if (substr($2,1,4)=="stx2") print $1,$3,$4}' stxdir/skesa_seqs > stxdir/stx2_skesa_seqs;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
46 awk 'BEGIN { OFS="\t" }!seen[$1]++ {if (substr($2,1,4)=="stx1") print $1,$3,$4}' stxdir/duk_skesa_seqs > stxdir/dukstx1_skesa_seqs;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
47 awk 'BEGIN { OFS="\t" }!seen[$1]++ {if (substr($2,1,4)=="stx2") print $1,$3,$4}' stxdir/duk_skesa_seqs > stxdir/dukstx2_skesa_seqs;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
48 awk 'BEGIN { OFS="\t" }!seen[$1]++ {if (substr($2,1,4)=="stx1") print $1,$3,$4}' stxdir/spades_seqs > stxdir/stx1_spades_seqs;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
49 awk 'BEGIN { OFS="\t" }!seen[$1]++ {if (substr($2,1,4)=="stx2") print $1,$3,$4}' stxdir/spades_seqs > stxdir/stx2_spades_seqs;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
50 awk 'BEGIN { OFS="\t" }!seen[$1]++ {if (substr($2,1,4)=="stx1") print $1,$3,$4}' stxdir/duk_spades_seqs > stxdir/dukstx1_spades_seqs;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
51 awk 'BEGIN { OFS="\t" }!seen[$1]++ {if (substr($2,1,4)=="stx2") print $1,$3,$4}' stxdir/duk_spades_seqs > stxdir/dukstx2_spades_seqs;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
52 # CREATE COMBINED MULTIFASTA FROM SEQUENCES
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
53 perl $tooldir/scripts/MultifastaFromBlast.pl "stxdir/stx1_skesa_seqs,stxdir/dukstx1_skesa_seqs,stxdir/stx1_spades_seqs,stxdir/dukstx1_spades_seqs" "stxdir/multiassembly_stx1.fasta";
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
54 perl $tooldir/scripts/MultifastaFromBlast.pl "stxdir/stx2_skesa_seqs,stxdir/dukstx2_skesa_seqs,stxdir/stx2_spades_seqs,stxdir/dukstx2_spades_seqs" "stxdir/multiassembly_stx2.fasta";
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
55
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
56 # ALIGN AND GET CONSENSUS
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
57 stx1filesize=$(wc -c "stxdir/multiassembly_stx1.fasta" | awk '{print $1}');
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
58 if [ $stx1filesize -eq 0 ]
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
59 then
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
60 touch stxdir/multiassembly_stx1_consensus.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
61 else
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
62 cat $tooldir/data/stx1.fa >> stxdir/multiassembly_stx1.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
63 muscle -in stxdir/multiassembly_stx1.fasta -out stxdir/multiassembly_stx1_aligned.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
64 awk 'BEGIN {RS=">" ; ORS=""} substr($1,1,4)!="stx1" {print ">"$0}' stxdir/multiassembly_stx1_aligned.fasta > stxdir/multiassembly_stx1_aligned_clean.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
65 awk '/^>/ {printf("%s%s\n",(N>0?"\n":""),$0);N++;next;} {printf("%s",$0);} END {printf("\n");}' stxdir/multiassembly_stx1_aligned_clean.fasta > stxdir/multiassembly_stx1_aligned_linear.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
66 python $tooldir/scripts/GetConsensus.py -i stxdir/multiassembly_stx1_aligned_linear.fasta -o stxdir/multiassembly_stx1_consensus.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
67 fi
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
68
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
69 stx2filesize=$(wc -c "stxdir/multiassembly_stx2.fasta" | awk '{print $1}');
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
70 if [ $stx2filesize -eq 0 ]
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
71 then
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
72 touch stxdir/multiassembly_stx2_consensus.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
73 else
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
74 cat $tooldir/data/stx2.fa >> stxdir/multiassembly_stx2.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
75 muscle -in stxdir/multiassembly_stx2.fasta -out stxdir/multiassembly_stx2_aligned.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
76 awk 'BEGIN {RS=">" ; ORS=""} substr($1,1,4)!="stx2" {print ">"$0}' stxdir/multiassembly_stx2_aligned.fasta > stxdir/multiassembly_stx2_aligned_clean.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
77 awk '/^>/ {printf("%s%s\n",(N>0?"\n":""),$0);N++;next;} {printf("%s",$0);} END {printf("\n");}' stxdir/multiassembly_stx2_aligned_clean.fasta > stxdir/multiassembly_stx2_aligned_linear.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
78 python $tooldir/scripts/GetConsensus.py -i stxdir/multiassembly_stx2_aligned_linear.fasta -o stxdir/multiassembly_stx2_consensus.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
79 fi
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
80 cat stxdir/multiassembly_stx1_consensus.fasta > stx.fasta;
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
81 cat stxdir/multiassembly_stx2_consensus.fasta >> stx.fasta;