adding scripts for struct params

meta_data_analysis/dssp_df.py

@@ -0,0 +1,68 @@
+#!/home/tanu/anaconda3/envs/ContactMap/bin/python3
+# Read a DSSP file into a data frame and pretty-print it
+
+#https://jbloomlab.github.io/dms_tools2/dms_tools2.dssp.html
+import sys, os
+import pandas as pd
+import pprint as pp
+import dms_tools2
+import dms_tools2.dssp
+
+#%%
+# my working dir
+homedir = os.path.expanduser('~')  # spyder/python doesn't recognise tilde
+os.getcwd()
+os.chdir(homedir + '/git/LSHTM_analysis/meta_data_analysis/struct_params')
+os.getcwd()
+#%%
+# sample example
+dssp_file = "./3pl1.dssp"
+dssp_df = dms_tools2.dssp.processDSSP(dssp_file, chain='A')
+
+# outputs to console
+#returns df with ASA and RSA (base on Tien at al 2013 (theor.) values)
+#Link: https://en.wikipedia.org/wiki/Relative_accessible_surface_area
+pp.pprint(dssp_df)
+
+# write to csv
+dssp_df.to_csv('3pl1_dssp_df', header=True, index = False)
+
+#%% specify variables for input and output paths and filenames
+drug = "pyrazinamide"
+#gene = "pnca"
+
+datadir = homedir + "/git/Data"
+basedir =  datadir + "/" + drug + "/input"
+
+# input
+inpath = "/processed"
+in_filename  = "/3pl1.dssp"
+infile = basedir + inpath + in_filename
+#print(infile)
+
+# output file 
+outpath = "/output"
+outdir =   datadir + "/" + drug + outpath
+out_filename = "/3pl1_dssp_df"
+outfile =  outdir + out_filename
+print(outdir); print(outfile)
+
+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
+#%% <----- fixme
+dssp_file = infile
+dssp_df = dms_tools2.dssp.processDSSP(dssp_file, chain='A')
+
+#%%
+# write to csv
+dssp_df.to_csv(outfile, header=True, index = False)