lingyuzeng/analysis_pdb
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

add strip cut

Browse Source
This commit is contained in:
root
2024-01-24 17:14:02 +08:00
parent 2758bfb72f
commit 4cf2f808e5
1 changed files with 31 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

40
build_modeller.py
View File

@@ -86,15 +86,35 @@ class PDBModeler:
aln.append_model(mdl, align_codes=self.structure, atom_files=self.structure_file.as_posix()) aln.append_model(mdl, align_codes=self.structure, atom_files=self.structure_file.as_posix())
aln.append(file=self.ali_file.as_posix(), align_codes=self.sequence) aln.append(file=self.ali_file.as_posix(), align_codes=self.sequence)
aln.align2d() aln.align2d()
aln.write(file=(self.outdir / 'alignment.ali').as_posix(), alignment_format='PIR') aln.write(file=(self.outdir / 'alignment1.ali').as_posix(), alignment_format='PIR')
aln.write(file=(self.outdir / 'alignment.pap').as_posix(), alignment_format='PAP') aln.write(file=(self.outdir / 'alignment1.pap').as_posix(), alignment_format='PAP')
# save alignment in FASTA format
aln.write(file=(self.outdir / 'alignment1.fasta').as_posix(), alignment_format='FASTA')
slice_fasta = PDBModeler.align_sequences(self.outdir / 'alignment1.fasta')
slice_ali = self.outdir / 'alignment_slice.ali'
fx = pyfastx.Fasta(slice_fasta.as_posix(), build_index=True)
assert len(fx) == 1, "FASTA file should contain only one sequence"
PDBModeler.write_ali(slice_ali, fx[0].name, fx[0].seq)
env2 = Environ()
mdl2 = Model(env2, file=self.structure_file.as_posix(), model_segment=(f'FIRST:{self.chain}', f'LAST:{self.chain}'))
aln2 = Alignment(env2)
aln2.append_model(mdl2, align_codes=self.structure, atom_files=self.structure_file.as_posix())
aln2.append(file=slice_ali.as_posix(), align_codes=self.sequence)
aln2.align2d()
aln2.write(file=(self.outdir / 'alignment2.ali').as_posix(), alignment_format='PIR')
aln2.write(file=(self.outdir / 'alignment2.pap').as_posix(), alignment_format='PAP')
log.verbose() log.verbose()
# choose ali file
fix_ali_file = self.outdir / 'alignment2.ali' if (self.outdir / 'alignment2.ali').exists() else self.outdir / 'alignment1.ali'
env3 = Environ() env3 = Environ()
env3.io.atom_files_directory = ['.'] env3.io.atom_files_directory = ['.']
loop_model = LoopModel(env3, loop_model = LoopModel(env3,
alnfile=(self.outdir / 'alignment.ali').as_posix(), alnfile=fix_ali_file.as_posix(),
knowns=self.structure, knowns=self.structure,
sequence=self.sequence, sequence=self.sequence,
loop_assess_methods=(assess.DOPE, assess.GA341)) loop_assess_methods=(assess.DOPE, assess.GA341))
@@ -151,6 +171,13 @@ class PDBModeler:
if output.get('failure') is None: if output.get('failure') is None:
model_files.append(Path(output.get('name'))) model_files.append(Path(output.get('name')))
return model_files return model_files
@staticmethod
def write_ali(ali_file: Path, description: str, sequence: str):
with open(ali_file, 'w') as f:
f.write(f'>P1;{description}\n')
f.write(f'sequence:{description}:::::::0.00: 0.00\n')
f.write(f'{sequence}*')
def fasta_to_ali(self) -> Path: def fasta_to_ali(self) -> Path:
if not self.outdir.exists(): if not self.outdir.exists():
@@ -162,12 +189,7 @@ class PDBModeler:
fx = pyfastx.Fasta(self.fasta_file.as_posix(), build_index=True) fx = pyfastx.Fasta(self.fasta_file.as_posix(), build_index=True)
assert len(fx) == 1, "FASTA file should contain only one sequence" assert len(fx) == 1, "FASTA file should contain only one sequence"
PDBModeler.write_ali(ali_file, self.sequence, fx[0].seq)
with open(ali_file, 'w') as f:
f.write(f'>P1;{self.sequence}\n')
f.write(f'sequence:{self.sequence}:::::::0.00: 0.00\n')
f.write(f'{fx[0].seq}*')
return ali_file return ali_file
if __name__ == "__main__": if __name__ == "__main__":