Sitewise prediction

Interaction prediction

Protein residue sites involved in interaction are functionally important for cellular activities. They are generally defined as spatially closed residues in a protein complex. It would be easiest to obtain information of interaction sites from a protein complex structure per se. Thus, there are plenty of literature and work carried out to develop interaction predictors. To evaluate their predictive abilities, we implemented a powerful module pp.eval to evaluate sitewise predictors (include but not limited to interaction predictors).

We take protein 1aij chain L as an example to evaluate an interaction site predictor, tma300 in its transmembrane segments demarked by the PDBTM database. If you are not interested in transmembrane segments, you can set segment to whole.

Python

import pypropel as pp
import pandas as pd

prot_df = pd.DataFrame({
        'prot': ['1aij', ],
        'chain': ['L', ],
    })

pp.eval.sitewise_segment(
    prot_df=prot_df,
    dist_fp=to('data/pdb/complex/pdbtm/'),
    dist_limit=6.,
    tool_fp=to('data/isite/deeptminter/'),
    tool='tma300',
    pdb_fp=to('data/pdb/pdbtm/'),
    topo_fp=to('data/phobius/'),
    xml_fp=to('data/xml/'),
    fasta_fp=to('data/fasta/'),
    segment='pdbtm_tmh',
    sort=1,
    sv_fp=to('data/'),
)

Output

28/07/2024 22:48:41 logger: ================>Labeling data...
28/07/2024 22:48:41 logger: ================>Time to label distances 1aij L: 0.00400090217590332s.
=========>Segment lower pdb id: [  1  50 106 131 150 164 190 248 259 268]
=========>Segment upper pdb id: [ 31  89 115 149 163 171 231 258 267 281]
=========>Segment lower fasta id: [1, 50, 106, 131, 150, 164, 190, 248, 259, 268]
=========>Segment upper fasta id: [31, 89, 115, 149, 163, 171, 231, 258, 267, 281]
=========>Segment lower pdb id: [ 32  90 116 172 232]
=========>Segment upper pdb id: [ 49 105 130 189 247]
=========>Segment lower fasta id: [32, 90, 116, 172, 232]
=========>Segment upper fasta id: [49, 105, 130, 189, 247]
{'1aij_L': {83: 28, 42: 25, 17: 13, 8: 7}}

It results in 29 files for evaluation.

Fig 1. Files of multiple evaluation metrics