src/data_extract.py

import os
import json
import csv
import pandas as pd

def clean_value(v):
    if isinstance(v, str):
        return v.replace("\n", ",")
    if isinstance(v, list):
        return ";".join(map(str, v))
    return v


def torsion_stats_dict(file_path):
    try:
        df = pd.read_csv(file_path)
    except FileNotFoundError:
        return {}
    # Detect torsion columns automatically
    torsion_cols = [c for c in df.columns if "_phi" in c or "_psi" in c or "_omega" in c]

    grouped = df.groupby("cluster")

    count_df = grouped.size().rename("count")
    mean_df = grouped[torsion_cols].mean()
    std_df = grouped[torsion_cols].std()

    mean_df.columns = [c + "_mean" for c in mean_df.columns]
    std_df.columns = [c + "_std" for c in std_df.columns]

    result_df = pd.concat([count_df, mean_df, std_df], axis=1)

    # Convert dataframe → formatted dictionary
    result_dict = {}

    for cluster_id, row in result_df.iterrows():

        cluster_key = f"Cluster {int(cluster_id)}"
        cluster_data = {}

        for col, val in row.items():

            if col == "count":
                cluster_data[col] = int(val)

            else:
                cluster_data[col] = round(float(val), 2)

        result_dict[cluster_key] = cluster_data

    return result_dict


raw_loc='../raw_data/'

GS_list = [
    d for d in os.listdir(raw_loc)
    if "GS" in d and os.path.isdir(os.path.join(raw_loc, d))
]

print('GS_list',len(GS_list))

data_archetype_term_list = ["glytoucan", "ID", "name", "glycam", "iupac", "iupac_extended", "wurcs", "glycoct",
                            "smiles", "oxford", "mass", "motifs", "termini", "components", "composition", "rot_bonds",
                            "hbond_donor", "hbond_acceptor", "entropy", "clusters","coverage_clusters","silhouette_scores","coverage_clusters_per_main","pca_variance", "length", "package", "forcefield",
                            "temperature", "pressure", "salt", ]
data_alpha_term_list = ["glycam", "iupac", "iupac_extended", "glytoucan", "wurcs", ]
data_beta_term_list = ["glycam", "iupac", "iupac_extended", "glytoucan", "wurcs", ]
data_meta_term_list = ["common_names", "description", "keywords"]
glygen_term_list=["number_monosaccharides","species","classification","enzyme","crossref","mass_pme","tool_support","missing_score","glycan_type","byonic","gwb","motifs","subsumption","section_stats","history",]



# ---- Build header ----
header_list=[]
header_list.extend(['SNFG','torsion_table','torsion_analysis'])
header_list.extend(data_archetype_term_list)
header_list.extend(["a_"+i for i in data_alpha_term_list])
header_list.extend(["b_"+i for i in data_beta_term_list])
header_list.extend(data_meta_term_list)
header_list.extend(['calculated_torsion','a_calculated_torsion','b_calculated_torsion',])
header_list.extend(['glycosmos'])
header_list.extend(glygen_term_list)

output_file = "../data/glycoshape_data.csv"
csvfile=open(output_file, "w", newline="", encoding="utf-8")
writer = csv.writer(csvfile)


# ---- load calculated torsion ---
with open("../data/glycan_dictionary.json", "r", encoding="utf-8") as f:
    calculated_torsion = json.load(f)
    f.close()



# Write header
writer.writerow(header_list)

for GS in GS_list:
    #print(GS)
    GS_loc=raw_loc+GS+'/'
    row = []

    ### add snfg.svg
    v='./raw_data'+GS+'/'+'snfg.svg'
    row.append(clean_value(v))


    ### add raw torsion table
    v = './torsion_data/' + GS + '_torsion_data.txt'
    row.append(clean_value(v))

    ## analysis torsion
    v=torsion_stats_dict('.'+v)
    row.append(clean_value(v))

    ### load data.json
    with open(GS_loc+"data.json", "r", encoding="utf-8") as f:
        data = json.load(f)
        f.close()

    for i in data_archetype_term_list:
        v = data.get('archetype', {}).get(i, "")
        row.append(clean_value(v))

    for i in data_alpha_term_list:
        v = data.get('alpha', {}).get(i, "")
        row.append(clean_value(v))

    for i in data_beta_term_list:
        v = data.get('beta', {}).get(i, "")
        row.append(clean_value(v))

    for i in data_meta_term_list:
        v = data.get('search_meta', {}).get(i, "")
        row.append(clean_value(v))

    gtouch=data['archetype']['glytoucan']
    a_gtouch=data['alpha']['glytoucan']
    b_gtouch=data['beta']['glytoucan']
    print(GS,gtouch,a_gtouch,b_gtouch)

    ### load calculated torsion json
    try:
        v=calculated_torsion[gtouch]
    except KeyError:
        v={}
    row.append(clean_value(v))

    try:
        v = calculated_torsion[a_gtouch]
    except KeyError:
        v = {}
    row.append(clean_value(v))

    try:
        v = calculated_torsion[b_gtouch]
    except KeyError:
        v = {}
    row.append(clean_value(v))


    ### load glycosmos.json
    with open(GS_loc+"glycosmos.json", "r", encoding="utf-8") as f:
        glycosmos = json.load(f)
        f.close()

    #print(glycosmos)
    v = glycosmos
    row.append(clean_value(v))

    ### load glygen.json
    with open(GS_loc+"glygen.json", "r", encoding="utf-8") as f:
        glygen = json.load(f)
        f.close()

    for i in glygen_term_list:
        try:
            v = glygen[i]
        except :
            v=[]
        row.append(clean_value(v))

    writer.writerow(row)
    #print(header_list)
    #print(row)