Closed eric-jm-lang closed 4 years ago
Could you provide a PDB file of the input structure you are having this problem with as well as the complete command line you use to invoke CHAP? Without a concrete example, this problem may be hard to diagnose.
Have you already tried varying the -pm-pl-margin
parameter? This is the maximal distance a residue COG may have from the pore surface to be counted as pore-lining. Only pore-lining residues will be considered in the subsequent identification of pore-facing residues.
To be honnest, I didn't remember I posted that. I solved the problem by using the pore lining residues instead of the pore facing residues when ploting the channel radius profile.
i.e., I use pf = np.array(data["residueSummary"]["poreLining"]["mean"]) > 0.5
instead of pf = np.array(data["residueSummary"]["poreFacing"]["mean"]) > 0.5
One related thing is that the N- and C-term are capped (with Amber ACE and NHE residues), which led to an error because they are not recognised. Surprisingly, I still had the error when using the -hydrophob-fallback
option, not sure why. So I ended up adding those residues to the hydrophibicity scale json file. I am just mentioning this in case someone else face the same problem in the future.
It seems that I am facing a problem with the way the pore facing residues are selected by chap. Basically running the program on my protein returns a zero-array for the list of pore facing residues. To be able to generate a plot like the "radius_profile.png" example, I need to use the pore lining residues
data["residueSummary"]["poreFacing"]
but then I get most of the residues of my protein. I tried to change the selection string from "name CA" to "name CB" or include more atoms for example but without success. This is not much of a problem for my current project but I thought it would be worth mentioning that the way the pore facing residues are identified do not always work.