various changes

This commit is contained in:
Tanushree Tunstall 2021-02-09 14:42:44 +00:00
parent ca68996264
commit 56150ae3c8
12 changed files with 159 additions and 231 deletions

View file

@ -2,7 +2,7 @@ PDB=$1
n=$2 n=$2
OUTDIR=$3 OUTDIR=$3
cd ${OUTDIR} cd ${OUTDIR}
logger "Running mutrenamefiles with PDB: ${PDB} n: ${n} OUTDIR: ${OUTDIR}"
cp Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout Matrix_Hbonds_${PDB}_Repair_${n}_PN.txt cp Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout Matrix_Hbonds_${PDB}_Repair_${n}_PN.txt
sed -n '5,190p' Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout > Matrix_Hbonds_RR_${PDB}_Repair_${n}_PN.txt sed -n '5,190p' Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout > Matrix_Hbonds_RR_${PDB}_Repair_${n}_PN.txt
sed -n '194,379p' Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout > Matrix_Hbonds_MM_${PDB}_Repair_${n}_PN.txt sed -n '194,379p' Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout > Matrix_Hbonds_MM_${PDB}_Repair_${n}_PN.txt
@ -61,9 +61,3 @@ cp InteractingResidues_Volumetric_${PDB}_Repair_${n}_PN.fxout InteractingResidue
sed -i '1,5d' InteractingResidues_Volumetric_${PDB}_Repair_${n}_PN.txt sed -i '1,5d' InteractingResidues_Volumetric_${PDB}_Repair_${n}_PN.txt
cp InteractingResidues_Disulfide_${PDB}_Repair_${n}_PN.fxout InteractingResidues_Disulfide_${PDB}_Repair_${n}_PN.txt cp InteractingResidues_Disulfide_${PDB}_Repair_${n}_PN.fxout InteractingResidues_Disulfide_${PDB}_Repair_${n}_PN.txt
sed -i '1,5d' InteractingResidues_Disulfide_${PDB}_Repair_${n}_PN.txt sed -i '1,5d' InteractingResidues_Disulfide_${PDB}_Repair_${n}_PN.txt

View file

@ -1,68 +0,0 @@
PDB=$1
n=$2
#cd /home/tanu/git/LSHTM_analysis/foldx/
logger "Running mutrenamefiles_mac"
cp Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout Matrix_Hbonds_${PDB}_Repair_${n}_PN.txt
sed -n '5,190p' Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout > Matrix_Hbonds_RR_${PDB}_Repair_${n}_PN.txt
sed -n '194,379p' Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout > Matrix_Hbonds_MM_${PDB}_Repair_${n}_PN.txt
sed -n '383,568p' Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout > Matrix_Hbonds_SM_${PDB}_Repair_${n}_PN.txt
sed -n '572,757p' Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout > Matrix_Hbonds_SS_${PDB}_Repair_${n}_PN.txt
cp Matrix_Distances_${PDB}_Repair_${n}_PN.fxout Matrix_Distances_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,4d Matrix_Distances_${PDB}_Repair_${n}_PN.txt
cp Matrix_Volumetric_${PDB}_Repair_${n}_PN.fxout Matrix_Volumetric_${PDB}_Repair_${n}_PN.txt
sed -n '5,190p' Matrix_Volumetric_${PDB}_Repair_${n}_PN.fxout > Matrix_Volumetric_RR_${PDB}_Repair_${n}_PN.txt
sed -n '194,379p' Matrix_Volumetric_${PDB}_Repair_${n}_PN.fxout > Matrix_Volumetric_MM_${PDB}_Repair_${n}_PN.txt
sed -n '383,568p' Matrix_Volumetric_${PDB}_Repair_${n}_PN.fxout > Matrix_Volumetric_SM_${PDB}_Repair_${n}_PN.txt
sed -n '572,757p' Matrix_Volumetric_${PDB}_Repair_${n}_PN.fxout > Matrix_Volumetric_SS_${PDB}_Repair_${n}_PN.txt
cp Matrix_Electro_${PDB}_Repair_${n}_PN.fxout Matrix_Electro_${PDB}_Repair_${n}_PN.txt
sed -n '5,190p' Matrix_Electro_${PDB}_Repair_${n}_PN.fxout > Matrix_Electro_RR_${PDB}_Repair_${n}_PN.txt
sed -n '194,379p' Matrix_Electro_${PDB}_Repair_${n}_PN.fxout > Matrix_Electro_MM_${PDB}_Repair_${n}_PN.txt
sed -n '383,568p' Matrix_Electro_${PDB}_Repair_${n}_PN.fxout > Matrix_Electro_SM_${PDB}_Repair_${n}_PN.txt
sed -n '572,757p' Matrix_Electro_${PDB}_Repair_${n}_PN.fxout > Matrix_Electro_SS_${PDB}_Repair_${n}_PN.txt
cp Matrix_Disulfide_${PDB}_Repair_${n}_PN.fxout Matrix_Disulfide_${PDB}_Repair_${n}_PN.txt
sed -n '5,190p' Matrix_Disulfide_${PDB}_Repair_${n}_PN.fxout > Matrix_Disulfide_RR_${PDB}_Repair_${n}_PN.txt
sed -n '194,379p' Matrix_Disulfide_${PDB}_Repair_${n}_PN.fxout > Matrix_Disulfide_MM_${PDB}_Repair_${n}_PN.txt
sed -n '383,568p' Matrix_Disulfide_${PDB}_Repair_${n}_PN.fxout > Matrix_Disulfide_SM_${PDB}_Repair_${n}_PN.txt
sed -n '572,757p' Matrix_Disulfide_${PDB}_Repair_${n}_PN.fxout > Matrix_Disulfide_SS_${PDB}_Repair_${n}_PN.txt
cp Matrix_Partcov_${PDB}_Repair_${n}_PN.fxout Matrix_Partcov_${PDB}_Repair_${n}_PN.txt
sed -n '5,190p' Matrix_Partcov_${PDB}_Repair_${n}_PN.fxout > Matrix_Partcov_RR_${PDB}_Repair_${n}_PN.txt
sed -n '194,379p' Matrix_Partcov_${PDB}_Repair_${n}_PN.fxout > Matrix_Partcov_MM_${PDB}_Repair_${n}_PN.txt
sed -n '383,568p' Matrix_Partcov_${PDB}_Repair_${n}_PN.fxout > Matrix_Partcov_SM_${PDB}_Repair_${n}_PN.txt
sed -n '572,757p' Matrix_Partcov_${PDB}_Repair_${n}_PN.fxout > Matrix_Partcov_SS_${PDB}_Repair_${n}_PN.txt
cp Matrix_VdWClashes_${PDB}_Repair_${n}_PN.fxout Matrix_VdWClashes_${PDB}_Repair_${n}_PN.txt
sed -n '5,190p' Matrix_VdWClashes_${PDB}_Repair_${n}_PN.fxout > Matrix_VdWClashes_RR_${PDB}_Repair_${n}_PN.txt
sed -n '194,379p' Matrix_VdWClashes_${PDB}_Repair_${n}_PN.fxout > Matrix_VdWClashes_MM_${PDB}_Repair_${n}_PN.txt
sed -n '383,568p' Matrix_VdWClashes_${PDB}_Repair_${n}_PN.fxout > Matrix_VdWClashes_SM_${PDB}_Repair_${n}_PN.txt
sed -n '572,757p' Matrix_VdWClashes_${PDB}_Repair_${n}_PN.fxout > Matrix_VdWClashes_SS_${PDB}_Repair_${n}_PN.txt
cp AllAtoms_Disulfide_${PDB}_Repair_${n}_PN.fxout AllAtoms_Disulfide_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,2d AllAtoms_Disulfide_${PDB}_Repair_${n}_PN.txt
cp AllAtoms_Electro_${PDB}_Repair_${n}_PN.fxout AllAtoms_Electro_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,2d AllAtoms_Electro_${PDB}_Repair_${n}_PN.txt
cp AllAtoms_Hbonds_${PDB}_Repair_${n}_PN.fxout AllAtoms_Hbonds_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,2d AllAtoms_Hbonds_${PDB}_Repair_${n}_PN.txt
cp AllAtoms_Partcov_${PDB}_Repair_${n}_PN.fxout AllAtoms_Partcov_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,2d AllAtoms_Partcov_${PDB}_Repair_${n}_PN.txt
cp AllAtoms_VdWClashes_${PDB}_Repair_${n}_PN.fxout AllAtoms_VdWClashes_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,2d AllAtoms_VdWClashes_${PDB}_Repair_${n}_PN.txt
cp AllAtoms_Volumetric_${PDB}_Repair_${n}_PN.fxout AllAtoms_Volumetric_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,2d AllAtoms_Volumetric_${PDB}_Repair_${n}_PN.txt
cp InteractingResidues_VdWClashes_${PDB}_Repair_${n}_PN.fxout InteractingResidues_VdWClashes_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,5d InteractingResidues_VdWClashes_${PDB}_Repair_${n}_PN.txt
cp InteractingResidues_Distances_${PDB}_Repair_${n}_PN.fxout InteractingResidues_Distances_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,5d InteractingResidues_Distances_${PDB}_Repair_${n}_PN.txt
cp InteractingResidues_Electro_${PDB}_Repair_${n}_PN.fxout InteractingResidues_Electro_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,5d InteractingResidues_Electro_${PDB}_Repair_${n}_PN.txt
cp InteractingResidues_Hbonds_${PDB}_Repair_${n}_PN.fxout InteractingResidues_Hbonds_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,5d InteractingResidues_Hbonds_${PDB}_Repair_${n}_PN.txt
cp InteractingResidues_Partcov_${PDB}_Repair_${n}_PN.fxout InteractingResidues_Partcov_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,5d InteractingResidues_Partcov_${PDB}_Repair_${n}_PN.txt
cp InteractingResidues_Volumetric_${PDB}_Repair_${n}_PN.fxout InteractingResidues_Volumetric_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,5d InteractingResidues_Volumetric_${PDB}_Repair_${n}_PN.txt
cp InteractingResidues_Disulfide_${PDB}_Repair_${n}_PN.fxout InteractingResidues_Disulfide_${PDB}_Repair_${n}_PN.txt
sed -i .bak -e 1,5d InteractingResidues_Disulfide_${PDB}_Repair_${n}_PN.txt

View file

@ -1,10 +0,0 @@
PDB=$1
A=$2
B=$3
n=$4
OUTDIR=$5
cd ${OUTDIR}
logger "Running mutruncomplex"
foldx --command=AnalyseComplex --pdb="${PDB}_Repair_${n}.pdb" --analyseComplexChains=${A},${B} --water=PREDICT --vdwDesign=1
cp ${OUTDIR}/Summary_${PDB}_Repair_${n}_AC.fxout ${OUTDIR}/Summary_${PDB}_Repair_${n}_AC.txt
#sed -i .bak -e 1,8d ${OUTDIR}/Summary_${PDB}_Repair_${n}_AC.txt

View file

@ -1,7 +1,7 @@
PDB=$1 PDB=$1
OUTDIR=$2 OUTDIR=$2
cd ${OUTDIR} cd ${OUTDIR}
logger "Running renamefiles"
cp Dif_${PDB}_Repair.fxout Dif_${PDB}_Repair.txt cp Dif_${PDB}_Repair.fxout Dif_${PDB}_Repair.txt
sed -i '1,8d' Dif_${PDB}_Repair.txt sed -i '1,8d' Dif_${PDB}_Repair.txt
cp Matrix_Hbonds_${PDB}_Repair_PN.fxout Matrix_Hbonds_${PDB}_Repair_PN.txt cp Matrix_Hbonds_${PDB}_Repair_PN.fxout Matrix_Hbonds_${PDB}_Repair_PN.txt
@ -62,9 +62,3 @@ cp InteractingResidues_Volumetric_${PDB}_Repair_PN.fxout InteractingResidues_Vol
sed -i '1,5d' InteractingResidues_Volumetric_${PDB}_Repair_PN.txt sed -i '1,5d' InteractingResidues_Volumetric_${PDB}_Repair_PN.txt
cp InteractingResidues_Disulfide_${PDB}_Repair_PN.fxout InteractingResidues_Disulfide_${PDB}_Repair_PN.txt cp InteractingResidues_Disulfide_${PDB}_Repair_PN.fxout InteractingResidues_Disulfide_${PDB}_Repair_PN.txt
sed -i '1,5d' InteractingResidues_Disulfide_${PDB}_Repair_PN.txt sed -i '1,5d' InteractingResidues_Disulfide_${PDB}_Repair_PN.txt

View file

@ -1,68 +0,0 @@
PDB=$1
logger "Running renamefiles_mac"
#cp Dif_${PDB}_Repair.fxout Dif_${PDB}_Repair.txt
sed -i '.bak' -e 1,8d Dif_${PDB}_Repair.txt
cp Matrix_Hbonds_${PDB}_Repair_PN.fxout Matrix_Hbonds_${PDB}_Repair_PN.txt
sed -n '5,190p' Matrix_Hbonds_${PDB}_Repair_PN.fxout > Matrix_Hbonds_RR_${PDB}_Repair_PN.txt
sed -n '194,379p' Matrix_Hbonds_${PDB}_Repair_PN.fxout > Matrix_Hbonds_MM_${PDB}_Repair_PN.txt
sed -n '383,568p' Matrix_Hbonds_${PDB}_Repair_PN.fxout > Matrix_Hbonds_SM_${PDB}_Repair_PN.txt
sed -n '572,757p' Matrix_Hbonds_${PDB}_Repair_PN.fxout > Matrix_Hbonds_SS_${PDB}_Repair_PN.txt
cp Matrix_Distances_${PDB}_Repair_PN.fxout Matrix_Distances_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,4d Matrix_Distances_${PDB}_Repair_PN.txt
cp Matrix_Volumetric_${PDB}_Repair_PN.fxout Matrix_Volumetric_${PDB}_Repair_PN.txt
sed -n '5,190p' Matrix_Volumetric_${PDB}_Repair_PN.fxout > Matrix_Volumetric_RR_${PDB}_Repair_PN.txt
sed -n '194,379p' Matrix_Volumetric_${PDB}_Repair_PN.fxout > Matrix_Volumetric_MM_${PDB}_Repair_PN.txt
sed -n '383,568p' Matrix_Volumetric_${PDB}_Repair_PN.fxout > Matrix_Volumetric_SM_${PDB}_Repair_PN.txt
sed -n '572,757p' Matrix_Volumetric_${PDB}_Repair_PN.fxout > Matrix_Volumetric_SS_${PDB}_Repair_PN.txt
cp Matrix_Electro_${PDB}_Repair_PN.fxout Matrix_Electro_${PDB}_Repair_PN.txt
sed -n '5,190p' Matrix_Electro_${PDB}_Repair_PN.fxout > Matrix_Electro_RR_${PDB}_Repair_PN.txt
sed -n '194,379p' Matrix_Electro_${PDB}_Repair_PN.fxout > Matrix_Electro_MM_${PDB}_Repair_PN.txt
sed -n '383,568p' Matrix_Electro_${PDB}_Repair_PN.fxout > Matrix_Electro_SM_${PDB}_Repair_PN.txt
sed -n '572,757p' Matrix_Electro_${PDB}_Repair_PN.fxout > Matrix_Electro_SS_${PDB}_Repair_PN.txt
cp Matrix_Disulfide_${PDB}_Repair_PN.fxout Matrix_Disulfide_${PDB}_Repair_PN.txt
sed -n '5,190p' Matrix_Disulfide_${PDB}_Repair_PN.fxout > Matrix_Disulfide_RR_${PDB}_Repair_PN.txt
sed -n '194,379p' Matrix_Disulfide_${PDB}_Repair_PN.fxout > Matrix_Disulfide_MM_${PDB}_Repair_PN.txt
sed -n '383,568p' Matrix_Disulfide_${PDB}_Repair_PN.fxout > Matrix_Disulfide_SM_${PDB}_Repair_PN.txt
sed -n '572,757p' Matrix_Disulfide_${PDB}_Repair_PN.fxout > Matrix_Disulfide_SS_${PDB}_Repair_PN.txt
cp Matrix_Partcov_${PDB}_Repair_PN.fxout Matrix_Partcov_${PDB}_Repair_PN.txt
sed -n '5,190p' Matrix_Partcov_${PDB}_Repair_PN.fxout > Matrix_Partcov_RR_${PDB}_Repair_PN.txt
sed -n '194,379p' Matrix_Partcov_${PDB}_Repair_PN.fxout > Matrix_Partcov_MM_${PDB}_Repair_PN.txt
sed -n '383,568p' Matrix_Partcov_${PDB}_Repair_PN.fxout > Matrix_Partcov_SM_${PDB}_Repair_PN.txt
sed -n '572,757p' Matrix_Partcov_${PDB}_Repair_PN.fxout > Matrix_Partcov_SS_${PDB}_Repair_PN.txt
cp Matrix_VdWClashes_${PDB}_Repair_PN.fxout Matrix_VdWClashes_${PDB}_Repair_PN.txt
sed -n '5,190p' Matrix_VdWClashes_${PDB}_Repair_PN.fxout > Matrix_VdWClashes_RR_${PDB}_Repair_PN.txt
sed -n '194,379p' Matrix_VdWClashes_${PDB}_Repair_PN.fxout > Matrix_VdWClashes_MM_${PDB}_Repair_PN.txt
sed -n '383,568p' Matrix_VdWClashes_${PDB}_Repair_PN.fxout > Matrix_VdWClashes_SM_${PDB}_Repair_PN.txt
sed -n '572,757p' Matrix_VdWClashes_${PDB}_Repair_PN.fxout > Matrix_VdWClashes_SS_${PDB}_Repair_PN.txt
cp AllAtoms_Disulfide_${PDB}_Repair_PN.fxout AllAtoms_Disulfide_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,2d AllAtoms_Disulfide_${PDB}_Repair_PN.txt
cp AllAtoms_Electro_${PDB}_Repair_PN.fxout AllAtoms_Electro_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,2d AllAtoms_Electro_${PDB}_Repair_PN.txt
cp AllAtoms_Hbonds_${PDB}_Repair_PN.fxout AllAtoms_Hbonds_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,2d AllAtoms_Hbonds_${PDB}_Repair_PN.txt
cp AllAtoms_Partcov_${PDB}_Repair_PN.fxout AllAtoms_Partcov_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,2d AllAtoms_Partcov_${PDB}_Repair_PN.txt
cp AllAtoms_VdWClashes_${PDB}_Repair_PN.fxout AllAtoms_VdWClashes_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,2d AllAtoms_VdWClashes_${PDB}_Repair_PN.txt
cp AllAtoms_Volumetric_${PDB}_Repair_PN.fxout AllAtoms_Volumetric_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,2d AllAtoms_Volumetric_${PDB}_Repair_PN.txt
cp InteractingResidues_VdWClashes_${PDB}_Repair_PN.fxout InteractingResidues_VdWClashes_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,5d InteractingResidues_VdWClashes_${PDB}_Repair_PN.txt
cp InteractingResidues_Distances_${PDB}_Repair_PN.fxout InteractingResidues_Distances_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,5d InteractingResidues_Distances_${PDB}_Repair_PN.txt
cp InteractingResidues_Electro_${PDB}_Repair_PN.fxout InteractingResidues_Electro_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,5d InteractingResidues_Electro_${PDB}_Repair_PN.txt
cp InteractingResidues_Hbonds_${PDB}_Repair_PN.fxout InteractingResidues_Hbonds_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,5d InteractingResidues_Hbonds_${PDB}_Repair_PN.txt
cp InteractingResidues_Partcov_${PDB}_Repair_PN.fxout InteractingResidues_Partcov_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,5d InteractingResidues_Partcov_${PDB}_Repair_PN.txt
cp InteractingResidues_Volumetric_${PDB}_Repair_PN.fxout InteractingResidues_Volumetric_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,5d InteractingResidues_Volumetric_${PDB}_Repair_PN.txt
cp InteractingResidues_Disulfide_${PDB}_Repair_PN.fxout InteractingResidues_Disulfide_${PDB}_Repair_PN.txt
sed -i '.bak' -e 1,5d InteractingResidues_Disulfide_${PDB}_Repair_PN.txt

View file

@ -1,9 +0,0 @@
INDIR=$1
PDB=$2
OUTDIR=$3
logger "Running repairPDB"
#foldx --command=RepairPDB --pdb="${PDB}.pdb" --ionStrength=0.05 --pH=7 --water=PREDICT --vdwDesign=1 outPDB=true --output-dir=${OUTDIR}
foldx --command=RepairPDB --pdb-dir=${INDIR} --pdb=${PDB} --ionStrength=0.05 --pH=7 --water=PREDICT --vdwDesign=1 outPDB=true --output-dir=${OUTDIR}

View file

@ -1,6 +1,7 @@
#!/usr/bin/env python3 #!/usr/bin/env python3
import subprocess import subprocess
import os import os
import sys
import numpy as np import numpy as np
import pandas as pd import pandas as pd
from contextlib import suppress from contextlib import suppress
@ -8,6 +9,8 @@ from pathlib import Path
import re import re
import csv import csv
import argparse import argparse
import shutil
import time
#https://realpython.com/python-pathlib/ #https://realpython.com/python-pathlib/
# FIXME # FIXME
@ -22,8 +25,8 @@ homedir = os.path.expanduser('~')
# set working dir # set working dir
os.getcwd() os.getcwd()
os.chdir(homedir + '/git/LSHTM_analysis/foldx/') #os.chdir(homedir + '/git/LSHTM_analysis/foldx/')
os.getcwd() #os.getcwd()
#======================================================================= #=======================================================================
#%% command line args #%% command line args
@ -37,7 +40,7 @@ arg_parser.add_argument('-i', '--input_dir', help = 'Input dir containing pdb fi
arg_parser.add_argument('-o', '--output_dir', help = 'Output dir for results. By default, it assmes homedir + <drug> + output') arg_parser.add_argument('-o', '--output_dir', help = 'Output dir for results. By default, it assmes homedir + <drug> + output')
arg_parser.add_argument('-p', '--process_dir', help = 'Temp processing dir for running foldX. By default, it assmes homedir + <drug> + processing. Make sure it is somewhere with LOTS of storage as it writes all output!') #FIXME arg_parser.add_argument('-p', '--process_dir', help = 'Temp processing dir for running foldX. By default, it assmes homedir + <drug> + processing. Make sure it is somewhere with LOTS of storage as it writes all output!') #FIXME
arg_parser.add_argument('-pdb', '--pdb_file', help = 'PDB File to process. By default, it assmumes a file called <gene>_complex.pdb in input_dir') arg_parser.add_argument('-pdb', '--pdb_file', help = 'PDB File to process. By default, it assmumes a file called <gene>_complex.pdb in input_dir') # DO NOT specify an absolute path
arg_parser.add_argument('-m', '--mutation_file', help = 'Mutation list. By default, assumes a file called <gene>_mcsm_snps.csv exists') arg_parser.add_argument('-m', '--mutation_file', help = 'Mutation list. By default, assumes a file called <gene>_mcsm_snps.csv exists')
# FIXME: Doesn't work with 2 chains yet! # FIXME: Doesn't work with 2 chains yet!
@ -77,34 +80,45 @@ if not datadir:
datadir = homedir + '/' + 'git/Data' datadir = homedir + '/' + 'git/Data'
if not indir: if not indir:
indir = datadir + '/' + drug + '/input' indir =
datadir + '/' + drug + '/input'
if not outdir: if not outdir:
outdir = datadir + '/' + drug + '/output' outdir = datadir + '/' + drug + '/output'
#TODO: perhaps better handled by refactoring code to prevent generating lots of output files! #TODO: perhaps better handled by refactoring code to prevent generating lots of output files!
if not process_dir: if not process_dir:
process_dir = datadir + '/' + drug +'/' + 'processing' process_dir = datadir + '/' + drug + '/processing'
# Make all paths absolute in case the user forgot
indir = os.path.abspath(indir)
process_dir = os.path.abspath(process_dir)
outdir = os.path.abspath(outdir)
datadir = os.path.abspath(datadir)
#======= #=======
# input # input
#======= #=======
# FIXME # FIXME
if pdb_filename: if pdb_filename:
pdb_filename = os.path.abspath(pdb_filename)
pdb_name = Path(pdb_filename).stem pdb_name = Path(pdb_filename).stem
infile_pdb = pdb_filename
else: else:
pdb_filename = gene.lower() + '_complex.pdb' pdb_filename = gene.lower() + '_complex.pdb'
pdb_name = Path(pdb_filename).stem pdb_name = Path(pdb_filename).stem
infile_pdb = indir + '/' + pdb_filename
infile_pdb = indir + '/' + pdb_filename
actual_pdb_filename = Path(infile_pdb).name actual_pdb_filename = Path(infile_pdb).name
if mut_filename: if mut_filename:
mutation_file = mut_filename mutation_file = os.path.abspath(mut_filename)
infile_muts = mutation_file
print('User-provided mutation file in use:', infile_muts)
else: else:
mutation_file = gene.lower() + '_mcsm_formatted_snps.csv' mutation_file = gene.lower() + '_mcsm_formatted_snps.csv'
infile_muts = outdir + '/' + mutation_file
infile_muts = outdir + '/' + mutation_file print('WARNING: Assuming default mutation file:', infile_muts)
#======= #=======
# output # output
@ -116,6 +130,7 @@ print('Arguments being passed:'
, '\nDrug:', args.drug , '\nDrug:', args.drug
, '\ngene:', args.gene , '\ngene:', args.gene
, '\ninput dir:', indir , '\ninput dir:', indir
, '\nprocess dir:', process_dir
, '\noutput dir:', outdir , '\noutput dir:', outdir
, '\npdb file:', infile_pdb , '\npdb file:', infile_pdb
, '\npdb name:', pdb_name , '\npdb name:', pdb_name
@ -124,6 +139,9 @@ print('Arguments being passed:'
, '\nchain1:', args.chain1 , '\nchain1:', args.chain1
, '\noutput file:', outfile_foldx , '\noutput file:', outfile_foldx
, '\n=============================================================') , '\n=============================================================')
#### Delay for 10 seconds to check the params ####
time.sleep(10)
#======================================================================= #=======================================================================
def getInteractionEnergy(filename): def getInteractionEnergy(filename):
@ -184,6 +202,19 @@ def loadFiles(df):
f.close() f.close()
return np.asarray(resultList, dtype=np.float32) return np.asarray(resultList, dtype=np.float32)
# TODO: put the subprocess call in a 'def'
#def repairPDB():
# subprocess.call(['foldx'
# , '--command=RepairPDB'
# , '--pdb-dir=' + indir
# , '--pdb=' + actual_pdb_filename
# , '--ionStrength=0.05'#
# , '--pH=7'
# , '--water=PREDICT'
# , '--vdwDesign=1'
# , 'outPDB=true'
# , '--output-dir=' + process_dir])
#======================================================================= #=======================================================================
def main(): def main():
pdbname = pdb_name pdbname = pdb_name
@ -195,37 +226,128 @@ def main():
nmuts = len(mutlist) nmuts = len(mutlist)
print(nmuts) print(nmuts)
print(mutlist) print(mutlist)
print('start') print('start')
#subprocess.check_output(['bash','repairPDB.sh', pdbname, process_dir]) # some common parameters for foldX
subprocess.check_output(['bash','repairPDB.sh', indir, actual_pdb_filename, process_dir]) foldx_common=' --ionStrength=0.05 --pH=7 --water=PREDICT --vdwDesign=1 '
print('end') print('\033[95mSTAGE: repair PDB (foldx subprocess) \033[0m')
output = subprocess.check_output(['bash', 'runfoldx.sh', pdbname, process_dir]) print('Running foldx RepairPDB for WT')
subprocess.call(['foldx'
, '--command=RepairPDB'
, foldx_common
, '--pdb-dir=' + indir
, '--pdb=' + actual_pdb_filename
, 'outPDB=true'
, '--output-dir=' + process_dir])
print('\033[95mCOMPLETED STAGE: repair PDB\033[0m')
print('\n==========================================================')
print('\033[95mSTAGE: Foldx commands BM, PN and SD (foldx subprocess) for WT\033[0m')
print('Running foldx BuildModel for WT')
subprocess.call(['foldx'
, '--command=BuildModel'
, foldx_common
, '--pdb-dir=' + process_dir
, '--pdb=' + pdbname + '_Repair.pdb'
, '--mutant-file="individual_list_' + pdbname +'.txt"'
, 'outPDB=true'
, '--numberOfRuns=1'
, '--output-dir=' + process_dir], cwd=process_dir)
print('Running foldx PrintNetworks for WT')
subprocess.call(['foldx'
, '--command=PrintNetworks'
, '--pdb-dir=' + process_dir
, '--pdb=' + pdbname + '_Repair.pdb'
, '--water=PREDICT'
, '--vdwDesign=1'
, '--output-dir=' + process_dir], cwd=process_dir)
print('Running foldx SequenceDetail for WT')
subprocess.call(['foldx'
, '--command=SequenceDetail'
, '--pdb-dir=' + process_dir
, '--pdb=' + pdbname + '_Repair.pdb'
, '--water=PREDICT'
, '--vdwDesign=1'
, '--output-dir=' + process_dir], cwd=process_dir)
print('\033[95mCOMPLETED STAGE: Foldx commands BM, PN and SD\033[0m')
print('\n==========================================================')
print('\033[95mSTAGE: Print Networks (foldx subprocess) for MT\033[0m')
for n in range(1,nmuts+1): for n in range(1,nmuts+1):
print(n) print('\033[95mNETWORK:\033[0m', n)
with suppress(Exception): print('Running foldx PrintNetworks for mutation', n)
subprocess.check_output(['bash', 'runPrintNetworks.sh', pdbname, str(n), process_dir]) subprocess.call(['foldx'
, '--command=PrintNetworks'
, '--pdb-dir=' + process_dir
, '--pdb=' + pdbname + '_Repair_' + str(n) + '.pdb'
, '--water=PREDICT'
, '--vdwDesign=1'
, '--output-dir=' + process_dir], cwd=process_dir)
print('\033[95mCOMPLETED STAGE: Print Networks (foldx subprocess) for MT\033[0m')
print('\n==========================================================')
print('\033[95mSTAGE: Rename Mutation Files (shell)\033[0m')
for n in range(1,nmuts+1): for n in range(1,nmuts+1):
print(n) print('\033[95mMUTATION:\033[0m', n)
print('\033[96mCommand:\033[0m mutrenamefiles.sh %s %s %s' % (pdbname, str(n), process_dir ))
#FIXME: bad design and needs to be done in a pythonic way
with suppress(Exception): with suppress(Exception):
subprocess.check_output(['bash', 'mutrenamefiles.sh', pdbname, str(n), process_dir]) subprocess.check_output(['bash', 'mutrenamefiles.sh', pdbname, str(n), process_dir])
print('\033[95mCOMPLETED STAGE: Rename Mutation Files (shell)\033[0m')
print('\n==========================================================')
print('\033[95mSTAGE: Rename Files (shell) for WT\033[0m')
# FIXME: this is bad design and needs to be done in a pythonic way
out = subprocess.check_output(['bash','renamefiles.sh', pdbname, process_dir]) out = subprocess.check_output(['bash','renamefiles.sh', pdbname, process_dir])
print('\033[95mCOMPLETED STAGE: Rename Files (shell) for WT\033[0m')
print('\n==========================================================')
if comp=='y': if comp=='y':
print('\033[95mSTAGE: Running foldx AnalyseComplex (foldx subprocess) for WT\033[0m')
chain1=chainA chain1=chainA
chain2=chainB chain2=chainB
with suppress(Exception): subprocess.call(['foldx'
subprocess.check_output(['bash','runcomplex.sh', pdbname, chain1, chain2, process_dir]) , '--command=AnalyseComplex'
for n in range(1,nmuts+1): , '--pdb-dir=' + process_dir
with suppress(Exception): , '--pdb=' + pdbname + '_Repair.pdb'
subprocess.check_output(['bash','mutruncomplex.sh', pdbname, chain1, chain2, str(n), process_dir]) , '--analyseComplexChains=' + chain1 + ',' + chain2
, '--water=PREDICT'
, '--vdwDesign=1'
, '--output-dir=' + process_dir], cwd=process_dir)
interactions = ['Distances','Electro_RR','Electro_MM','Electro_SM','Electro_SS','Disulfide_RR','Disulfide_MM','Disulfide_SM','Disulfide_SS', # FIXME why would we ever need to do this?!? Cargo-culted from runcomplex.sh
'Hbonds_RR','Hbonds_MM','Hbonds_SM','Hbonds_SS','Partcov_RR','Partcov_MM','Partcov_SM','Partcov_SS','VdWClashes_RR','VdWClashes_MM', ac_source = process_dir + '/Summary_' + pdbname + '_Repair_AC.fxout'
'VdWClashes_SM','VdWClashes_SS','Volumetric_RR','Volumetric_MM','Volumetric_SM','Volumetric_SS'] ac_dest = process_dir + '/Summary_' + pdbname + '_Repair_AC.txt'
shutil.copyfile(ac_source, ac_dest)
print('\033[95mCOMPLETED STAGE: foldx AnalyseComplex (subprocess) for WT:\033[0m', n)
for n in range(1,nmuts+1):
print('\033[95mSTAGE: Running foldx AnalyseComplex (foldx subprocess) for mutation:\033[0m', n)
subprocess.call(['foldx'
, '--command=AnalyseComplex'
, '--pdb-dir=' + process_dir
, '--pdb=' + pdbname + '_Repair_' + str(n) + '.pdb'
, '--analyseComplexChains=' + chain1 + ',' + chain2
, '--water=PREDICT'
, '--vdwDesign=1'
, '--output-dir=' + process_dir], cwd=process_dir)
# FIXME why would we ever need to do this?!? Cargo-culted from runcomplex.sh
ac_mut_source = process_dir + '/Summary_' + pdbname + '_Repair_' + str(n) +'_AC.fxout'
ac_mut_dest = process_dir + '/Summary_' + pdbname + '_Repair)' + str(n) +'_AC.txt'
shutil.copyfile(ac_mut_source, ac_mut_dest)
print('\033[95mCOMPLETED STAGE: foldx AnalyseComplex (subprocess) for mutation:\033[0m', n)
print('\n==========================================================')
interactions = ['Distances','Electro_RR','Electro_MM','Electro_SM','Electro_SS','Disulfide_RR','Disulfide_MM','Disulfide_SM','Disulfide_SS', 'Hbonds_RR','Hbonds_MM','Hbonds_SM','Hbonds_SS','Partcov_RR','Partcov_MM','Partcov_SM','Partcov_SS','VdWClashes_RR','VdWClashes_MM','VdWClashes_SM','VdWClashes_SS','Volumetric_RR','Volumetric_MM','Volumetric_SM','Volumetric_SS']
dGdatafile = process_dir + '/Dif_' + pdbname + '_Repair.txt' dGdatafile = process_dir + '/Dif_' + pdbname + '_Repair.txt'
dGdata = pd.read_csv(dGdatafile, sep = '\t') dGdata = pd.read_csv(dGdatafile, sep = '\t')
@ -267,8 +389,7 @@ def main():
print(d) print(d)
data[i+1] = d data[i+1] = d
interactions = ['ddG', 'Distances','Electro_RR','Electro_MM','Electro_SM','Electro_SS','Disulfide_RR','Disulfide_MM','Disulfide_SM','Disulfide_SS', 'Hbonds_RR','Hbonds_MM','Hbonds_SM','Hbonds_SS','Partcov_RR','Partcov_MM','Partcov_SM','Partcov_SS','VdWClashes_RR','VdWClashes_MM', interactions = ['ddG', 'Distances','Electro_RR','Electro_MM','Electro_SM','Electro_SS','Disulfide_RR','Disulfide_MM','Disulfide_SM','Disulfide_SS', 'Hbonds_RR','Hbonds_MM','Hbonds_SM','Hbonds_SS','Partcov_RR','Partcov_MM','Partcov_SM','Partcov_SS','VdWClashes_RR','VdWClashes_MM','VdWClashes_SM','VdWClashes_SS','Volumetric_RR','Volumetric_MM','Volumetric_SM','Volumetric_SS']
'VdWClashes_SM','VdWClashes_SS','Volumetric_RR','Volumetric_MM','Volumetric_SM','Volumetric_SS']
print(interactions) print(interactions)
@ -292,8 +413,7 @@ def main():
print(len(IE)) print(len(IE))
data = np.append(data,[IE], axis = 0) data = np.append(data,[IE], axis = 0)
print(data) print(data)
interactions = ['ddG','Distances','Electro_RR','Electro_MM','Electro_SM','Electro_SS','Disulfide_RR','Disulfide_MM','Disulfide_SM','Disulfide_SS', 'Hbonds_RR','Hbonds_MM','Hbonds_SM','Hbonds_SS','Partcov_RR','Partcov_MM','Partcov_SM','Partcov_SS','VdWClashes_RR','VdWClashes_MM', interactions = ['ddG','Distances','Electro_RR','Electro_MM','Electro_SM','Electro_SS','Disulfide_RR','Disulfide_MM','Disulfide_SM','Disulfide_SS','Hbonds_RR','Hbonds_MM','Hbonds_SM','Hbonds_SS','Partcov_RR','Partcov_MM','Partcov_SM','Partcov_SS','VdWClashes_RR','VdWClashes_MM','VdWClashes_SM','VdWClashes_SS','Volumetric_RR','Volumetric_MM','Volumetric_SM','Volumetric_SS','Interaction Energy']
'VdWClashes_SM','VdWClashes_SS','Volumetric_RR','Volumetric_MM','Volumetric_SM','Volumetric_SS','Interaction Energy']
mut_file = process_dir + '/individual_list_' + pdbname + '.txt' mut_file = process_dir + '/individual_list_' + pdbname + '.txt'
with open(mut_file) as csvfile: with open(mut_file) as csvfile:
@ -331,6 +451,7 @@ def main():
#outputfilename = 'foldx_results_' + pdbname + '.csv' #outputfilename = 'foldx_results_' + pdbname + '.csv'
#results.to_csv(outputfilename) #results.to_csv(outputfilename)
results2.to_csv(outputfilename, index = False) results2.to_csv(outputfilename, index = False)
print ('end')
if __name__ == '__main__': if __name__ == '__main__':
main() main()

View file

@ -1,7 +0,0 @@
PDB=$1
n=$2
OUTDIR=$3
logger "Running runPrintNetworks"
cd ${OUTDIR}
foldx --command=PrintNetworks --pdb="${PDB}_Repair_${n}.pdb" --water=PREDICT --vdwDesign=1 --output-dir=${OUTDIR}

View file

@ -1,9 +0,0 @@
PDB=$1
A=$2
B=$3
OUTDIR=$4
cd ${OUTDIR}
logger "Running runcomplex"
foldx --command=AnalyseComplex --pdb="${PDB}_Repair.pdb" --analyseComplexChains=${A},${B} --water=PREDICT --vdwDesign=1 --output-dir=${OUTDIR}
cp ${OUTDIR}/Summary_${PDB}_Repair_AC.fxout ${OUTDIR}/Summary_${PDB}_Repair_AC.txt
#sed -i .bak -e 1,8d ${OUTDIR}/Summary_${PDB}_Repair_AC.txt

View file

@ -1,9 +0,0 @@
PDB=$1
OUTDIR=$2
cd ${OUTDIR}
pwd
ls
logger "Running runfoldx"
foldx --command=BuildModel --pdb="${PDB}_Repair.pdb" --mutant-file="individual_list_${PDB}.txt" --ionStrength=0.05 --pH=7 --water=PREDICT --vdwDesign=1 --out-pdb=true --numberOfRuns=1 --output-dir=${OUTDIR}
foldx --command=PrintNetworks --pdb="${PDB}_Repair.pdb" --water=PREDICT --vdwDesign=1 --output-dir=${OUTDIR}
foldx --command=SequenceDetail --pdb="${PDB}_Repair.pdb" --water=PREDICT --vdwDesign=1 --output-dir=${OUTDIR}

View file

@ -2,7 +2,7 @@ PDB=$1
n=$2 n=$2
OUTDIR=$3 OUTDIR=$3
cd ${OUTDIR} cd ${OUTDIR}
#cd /home/git/LSHTM_analysis/foldx/test2
cp Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout Matrix_Hbonds_${PDB}_Repair_${n}_PN.txt cp Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout Matrix_Hbonds_${PDB}_Repair_${n}_PN.txt
sed -n '5,190p' Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout > Matrix_Hbonds_RR_${PDB}_Repair_${n}_PN.txt sed -n '5,190p' Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout > Matrix_Hbonds_RR_${PDB}_Repair_${n}_PN.txt
sed -n '194,379p' Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout > Matrix_Hbonds_MM_${PDB}_Repair_${n}_PN.txt sed -n '194,379p' Matrix_Hbonds_${PDB}_Repair_${n}_PN.fxout > Matrix_Hbonds_MM_${PDB}_Repair_${n}_PN.txt

View file

@ -1,8 +1,7 @@
PDB=$1 PDB=$1
OUTDIR=$2 OUTDIR=$2
cd ${OUTDIR} cd ${OUTDIR}
#cd /home/git/LSHTM_analysis/foldx/test2
#cd /home/git/LSHTM_analysis/foldx/test
cp Dif_${PDB}_Repair.fxout Dif_${PDB}_Repair.txt cp Dif_${PDB}_Repair.fxout Dif_${PDB}_Repair.txt
sed -i '1,8d' Dif_${PDB}_Repair.txt sed -i '1,8d' Dif_${PDB}_Repair.txt
cp Matrix_Hbonds_${PDB}_Repair_PN.fxout Matrix_Hbonds_${PDB}_Repair_PN.txt cp Matrix_Hbonds_${PDB}_Repair_PN.fxout Matrix_Hbonds_${PDB}_Repair_PN.txt