tidied and updated kd and dssp scripts & generated their respective outputs

meta_data_analysis/run_pdb_dssp.sh

@@ -1,51 +1,54 @@
-#!/usr/bin/env python3
-# -*- coding: utf-8 -*-
-"""
-Created on Sun Feb 16 11:21:44 2020
+#!/bin/bash
+#=======================================================================
+# Task: read a pdb file and generate a dssp output file
+# Input:
+# pdb file
 
-@author: tanu
-"""
-#!/usr/bin/bash
-# Run dssp exe
+# Output:
+# pdb_code.dssp
+# needs dssp exe on linux
+# more efficient to run dssp exe locally
 
-#%%
-# specify variables for input and output paths and filenames
-drug="pyrazinamide"
-#gene = "pnca"
+# note: double quotes for variable interpolation
+#=======================================================================
+# #%%specify variables for input and output paths and filenames
+drug='pyrazinamide'
+gene='pncA'
+#convert to lowercase for consistency in filenames
+gene_l=$(printf $gene | awk '{print tolower($0)}') 
+gene_match="${gene}_p."
+#printf "${gene_match}\n"
 
-datadir="~git/Data"
-basedir=${datadir}"/"${drug}
-echo${basedir}
+#==========
+# data dir
+#==========
+#indir = 'git/Data/pyrazinamide/input/original'
+#datadir="~git/Data"
+datadir="${HOME}/git/Data"
 
+#=======
 # input
-inpath="/original"
+#=======
+indir=${datadir}'/'${drug}'/input'
+printf "Input dir: ${indir}\n"
+in_filename='3pl1.pdb'
+#infile=${basedir}${inpath}${in_filename}
+infile=${indir}'/'${in_filename}
+printf "Input file: ${infile}\n"
 
-# uncomment as necessary
-in_filename="/3pl1.pdb"
+#=======
+# output 
+#=======
+outdir=${datadir}'/'${drug}$'/output'
+printf "Output dir: ${outdir}"
+#out_filename='3pl1.dssp'
+out_filename="${gene_l}.dssp"
+outfile=${outdir}'/'${out_filename}
+printf "Output file: ${outfile}\n"
 
-infile=${basedir}${inpath}${in_filename}
-echo${infile}
-
-# output file 
-outpath="/processed"
-outdir=${datadir}"/"${drug}${outpath}
-out_filename="/3pl1.dssp"
-outfile=${outdir}${out_filename}
-echo${outdir}
-
-if not os.path.exists(datadir):
-    print('Error!', datadir, 'does not exist. Please ensure it exists. Dir struc specified in README.md')
-    os.makedirs(datadir)
-    exit()
-
-if not os.path.exists(outdir):
-    print('Error!', outdir, 'does not exist.Please ensure it exists. Dir struc specified in README.md')
-    exit()
-    
-else:
-    print('Dir exists: Carrying on')
-# end of variable assignment for input and output files
-#%%
-# ommand line args
-dssp -i 3pl1.pdb -o 3pl1.dssp
-dssp -i 
+#%%end of variable assignment for input and output files
+#================================================================
+# command line arg to run dssp and create output file
+dssp -i ${infile} -o ${outfile}
+printf "Finished writing: ${outfile}\n"
+#printf "Filename: ${out_filename}\nlocated in: ${outdir}\n"