|
2
|
1 ## This is a sample PBS script for profiling STR from reference genome using STR-FM version 1.0.0 (April 20, 2014)
|
|
|
2 ##
|
|
|
3 ##requirement
|
|
|
4 ##1 reference genome in FASTA format --> ${INPUT}.fa
|
|
|
5 ##
|
|
|
6 echo " "
|
|
|
7 echo " "
|
|
|
8 echo "Job started on `hostname` at `date`"
|
|
|
9 cd /working/directory/
|
|
|
10 echo " "
|
|
|
11 echo " detect STR in reference genome" ## See detail in microsatellite.xml on https://github.com/Arkarachai/STR-FM
|
|
|
12 python microsatellite.py ${INPUT}.fa --fasta --period=1 --partialmotifs --minlength=4 --prefix=0 --suffix=0 --hamming=0 --multipleruns --flankdisplay=0 --splitbyvalidity >${INPUT}.mono.out
|
|
|
13 python microsatellite.py ${INPUT}.fa --fasta --period=2 --partialmotifs --minlength=6 --prefix=0 --suffix=0 --hamming=0 --multipleruns --flankdisplay=0 --splitbyvalidity >${INPUT}.di.out
|
|
|
14 python microsatellite.py ${INPUT}.fa --fasta --period=3 --partialmotifs --minlength=6 --prefix=0 --suffix=0 --hamming=0 --multipleruns --flankdisplay=0 --splitbyvalidity >${INPUT}.tri.out
|
|
|
15 python microsatellite.py ${INPUT}.fa --fasta --period=4 --partialmotifs --minlength=8 --prefix=0 --suffix=0 --hamming=0 --multipleruns --flankdisplay=0 --splitbyvalidity >${INPUT}.tetra.out
|
|
|
16
|
|
|
17 echo "formatting"
|
|
|
18 cat ${INPUT}.mono.out | awk 'BEGIN{FS="\t";OFS="\t"};{print $6,$2,$2+$1,$4,$1,length($4) }' > ${INPUT}.mono.TR
|
|
|
19 cat ${INPUT}.di.out | awk 'BEGIN{FS="\t";OFS="\t"};{print $6,$2,$2+$1,$4,$1,length($4) }' > ${INPUT}.di.TR
|
|
|
20 cat ${INPUT}.tri.out | awk 'BEGIN{FS="\t";OFS="\t"};{print $6,$2,$2+$1,$4,$1,length($4) }' > ${INPUT}.tri.TR
|
|
|
21 cat ${INPUT}.tetra.out | awk 'BEGIN{FS="\t";OFS="\t"};{print $6,$2,$2+$1,$4,$1,length($4) }' > ${INPUT}.tetra.TR
|
|
|
22
|
|
|
23
|
|
|
24
|
|
|
25 echo "Job end on `hostname` at `date`"
|
