shruthivis
commented
1 year ago - Line 9. fdr. help string needs to have FDR
- Current implementation is ok, but would be better to use inbuilt functions for parsing PDB (biopython), and calculating distance (e.g. numpy.linalg.norm). a. They are more efficient b. If PDB format changes you wouldnt have to deal with changing your code. c. You write a lot less code. d. You avoid errors by writing less code. E.g. that uses res1['CA'] -res2['CA'] has inbuilt distance calculation.
- For a small protein complex 5% FDR could be too much. Could make it less than 5%
- This code is misusing dictionaries in many ways. a. Storing object (particle) as a key to a dictionary is quite weird. Objects are memory heavy, so could quickly run into memory issues? More natural to store the key as residue number len(pdb_items.keys()) and values as x,y,z coords. b. Dictionaries are quite expensive. You only make them (instead of normal list) when you want to access by key which we are not doing here. We are simply using it in a for loop for which list would suffice. c. The line contents (ln.strip()) are not used at all. Instead of pdb_lines{particle:[ln_contents,residue_number] one could have pdb_residues[residue_number]=particle.
shreyas-arvindekar