Source code for miniworld.utils.hhpred_parser

import torch



def parse_segment(line_segment, query_ID,device):

    # remove only '\n'
    line_segment = [line.strip() for line in line_segment]
    line_segment = list(filter(None, line_segment))

    target_info = line_segment[1]
    target_ID = target_info.split()[0].replace(">","")
    match_info = line_segment[2]

    line_segment = line_segment[3:]
    Query_seq_lines = []
    Target_seq_lines = []

    seq_map = []
    query_seq = ""
    target_seq = ""
    confidence = ""

    for line in line_segment :
        if query_seq == "" and f"Q {query_ID}" in line:
            query_start = int(line.split()[2]) -1 
        if target_seq == "" and f"T {target_ID}" in line:
            target_start = int(line.split()[2]) -1

        if f"Q {query_ID}" in line:
            query_seq += line.split()[-3]
        elif f"T {target_ID}" in line:
            target_seq += line.split()[-3]
        elif f"Confidence" in line:
            confidence += line[22:]

    confidence = confidence.replace(" ", "0")

    query_idx = query_start
    target_idx = target_start

    query_map_idx = []
    target_map_idx = []
    confidence_tensor = []

    # Iterate through the aligned sequences to create the mapping
    for q_char, t_char, conf in zip(query_seq, target_seq, confidence):
        if q_char != '-' and t_char != '-':
            query_map_idx.append(query_idx)
            target_map_idx.append(target_idx)
            confidence_tensor.append(float(conf)/10)
            query_idx += 1
            target_idx += 1
        elif q_char == '-':
            target_idx += 1
        elif t_char == '-':
            query_idx += 1
    # Convert mapping to a tensor
    query_map_idx = torch.tensor(query_map_idx, device=device)
    target_map_idx = torch.tensor(target_map_idx, device=device)
    confidence = torch.tensor(confidence_tensor, device=device)

    return {
        "target_ID" : target_ID,
        "query_map_idx" : query_map_idx,
        "target_map_idx" : target_map_idx,
        "confidence" : confidence,
    }    





def hhpred_parser(hhpred_hhr_file, query_ID, device=torch.device('cpu')):
    with open(hhpred_hhr_file, 'r') as f:
        lines = f.readlines()

    line_seperated = []
    line_segment = []
    for line in lines:
        if line.startswith('No '):
            line_seperated.append(line_segment)
            line_segment = []
        line_segment.append(line)

    # we don't need first line_segment
    line_seperated = line_seperated[1:]

    parsed_info_list = []
    
    for line_segment in line_seperated:
        parsed_info = parse_segment(line_segment, query_ID, device)
        parsed_info_list.append(parsed_info)

    return parsed_info_list


if __name__ == "__main__":
    hhpred_hhr_file = "/home/casp16/run/TS.human/T1207/MiniWorld/templates/hhpred_6231556.hhr"
    query_ID = "T1207"
    hhpred_parser(hhpred_hhr_file, query_ID)