scripts/04_thermodynamic_analysis.py

import os
import json
import random
import numpy as np
import pandas as pd
from scipy import stats

RESULTS_DIR = "results"
SAMPLE_SIZE = 10000
SEED = 42

NN_PARAMS = {
    "AA": {"dH": -7.9,  "dS": -22.2},
    "TT": {"dH": -7.9,  "dS": -22.2},
    "AT": {"dH": -7.2,  "dS": -20.4},
    "TA": {"dH": -7.2,  "dS": -21.3},
    "CA": {"dH": -8.5,  "dS": -22.7},
    "AC": {"dH": -8.5,  "dS": -22.7},
    "GT": {"dH": -8.4,  "dS": -22.4},
    "TG": {"dH": -8.4,  "dS": -22.4},
    "CT": {"dH": -7.8,  "dS": -21.0},
    "TC": {"dH": -7.8,  "dS": -21.0},
    "GA": {"dH": -8.2,  "dS": -22.2},
    "AG": {"dH": -8.2,  "dS": -22.2},
    "CG": {"dH": -10.6, "dS": -27.2},
    "GC": {"dH": -9.8,  "dS": -24.4},
    "GG": {"dH": -8.0,  "dS": -19.9},
    "CC": {"dH": -8.0,  "dS": -19.9},
}

INIT_DH = 0.2
INIT_DS = -5.7
R = 1.987e-3
T_KELVIN = 310.15


def calc_melting_temp(seq):
    seq = seq.upper()
    dH = INIT_DH
    dS = INIT_DS
    gc = seq.count("G") + seq.count("C")
    if gc > 0:
        dH += 0.1
        dS -= 2.8
    else:
        dH += 2.2
        dS += 6.9
    for i in range(len(seq) - 1):
        dinuc = seq[i:i + 2]
        if dinuc in NN_PARAMS:
            dH += NN_PARAMS[dinuc]["dH"]
            dS += NN_PARAMS[dinuc]["dS"]
    dS_total = dS / 1000.0
    if abs(dS_total) < 1e-10:
        return None
    return dH / dS_total - 273.15


def calc_gibbs(seq, T=T_KELVIN):
    seq = seq.upper()
    dH = INIT_DH
    dS = INIT_DS
    for i in range(len(seq) - 1):
        dinuc = seq[i:i + 2]
        if dinuc in NN_PARAMS:
            dH += NN_PARAMS[dinuc]["dH"]
            dS += NN_PARAMS[dinuc]["dS"]
    return dH - T * (dS / 1000.0)


def random_sequence(length, rng):
    return "".join(rng.choice(["A", "C", "G", "T"]) for _ in range(length))


def secondary_structure_flags(seq):
    seq = seq.upper()
    hairpin = any(seq[i] == seq[-(i + 1)] for i in range(len(seq) // 2))
    return {"hairpin": hairpin, "g4": "GGGG" in seq, "imotif": "CCCC" in seq}


def main():
    nullomers_path = os.path.join(RESULTS_DIR, "nullomers_k11.txt")
    if not os.path.exists(nullomers_path):
        raise FileNotFoundError(f"{nullomers_path} not found. Run 01_nullomer_identification.py first.")

    with open(nullomers_path) as f:
        all_nullomers = [line.strip() for line in f if line.strip()]

    rng_np = np.random.default_rng(SEED)
    rng_py = random.Random(SEED)

    sample_size = min(SAMPLE_SIZE, len(all_nullomers))
    nullomer_sample = rng_np.choice(all_nullomers, size=sample_size, replace=False).tolist()
    k = len(nullomer_sample[0])
    random_seqs = [random_sequence(k, rng_py) for _ in range(sample_size)]

    records = []
    for seq in nullomer_sample:
        flags = secondary_structure_flags(seq)
        records.append({"sequence": seq, "group": "nullomer",
                         "Tm": calc_melting_temp(seq), "dG": calc_gibbs(seq),
                         "GC": (seq.count("G") + seq.count("C")) / len(seq), **flags})
    for seq in random_seqs:
        flags = secondary_structure_flags(seq)
        records.append({"sequence": seq, "group": "random",
                         "Tm": calc_melting_temp(seq), "dG": calc_gibbs(seq),
                         "GC": (seq.count("G") + seq.count("C")) / len(seq), **flags})

    df = pd.DataFrame(records).dropna(subset=["Tm", "dG"])
    df.to_csv(os.path.join(RESULTS_DIR, "nullomer_thermodynamics.csv"), index=False)

    null_df = df[df["group"] == "nullomer"]
    rand_df = df[df["group"] == "random"]

    t_tm, p_tm = stats.ttest_ind(null_df["Tm"], rand_df["Tm"])
    t_dg, p_dg = stats.ttest_ind(null_df["dG"], rand_df["dG"])
    r_gc_tm, p_gc_tm = stats.pearsonr(null_df["GC"], null_df["Tm"])
    delta_dg = null_df["dG"].mean() - rand_df["dG"].mean()
    boltzmann_fold = np.exp(-delta_dg / (R * T_KELVIN))

    summary = {
        "n_nullomers": int(len(null_df)),
        "n_random": int(len(rand_df)),
        "nullomer_Tm_mean": round(float(null_df["Tm"].mean()), 2),
        "nullomer_Tm_std": round(float(null_df["Tm"].std()), 2),
        "random_Tm_mean": round(float(rand_df["Tm"].mean()), 2),
        "random_Tm_std": round(float(rand_df["Tm"].std()), 2),
        "Tm_difference": round(float(null_df["Tm"].mean() - rand_df["Tm"].mean()), 2),
        "Tm_t_statistic": round(float(t_tm), 3),
        "Tm_p_value": float(p_tm),
        "nullomer_dG_mean": round(float(null_df["dG"].mean()), 2),
        "nullomer_dG_std": round(float(null_df["dG"].std()), 2),
        "random_dG_mean": round(float(rand_df["dG"].mean()), 2),
        "random_dG_std": round(float(rand_df["dG"].std()), 2),
        "delta_dG_kcal_mol": round(float(delta_dg), 2),
        "boltzmann_fold_disadvantage": round(float(boltzmann_fold), 1),
        "dG_t_statistic": round(float(t_dg), 3),
        "dG_p_value": float(p_dg),
        "GC_Tm_pearson_r": round(float(r_gc_tm), 3),
        "GC_Tm_p_value": float(p_gc_tm),
        "pct_very_stable_nullomers": round(float((null_df["dG"] < -10).mean() * 100), 1),
        "pct_very_stable_random": round(float((rand_df["dG"] < -10).mean() * 100), 1),
        "pct_hairpin": round(float(null_df["hairpin"].mean() * 100), 1),
        "pct_g4": round(float(null_df["g4"].mean() * 100), 1),
        "pct_imotif": round(float(null_df["imotif"].mean() * 100), 1),
    }

    with open(os.path.join(RESULTS_DIR, "thermodynamic_summary.json"), "w") as f:
        json.dump(summary, f, indent=2)

    print(f"Tm: {summary['nullomer_Tm_mean']} vs {summary['random_Tm_mean']} °C  "
          f"t={summary['Tm_t_statistic']}  p={summary['Tm_p_value']:.2e}\n"
          f"dG: {summary['nullomer_dG_mean']} vs {summary['random_dG_mean']} kcal/mol  "
          f"ΔΔG={summary['delta_dG_kcal_mol']}  {summary['boltzmann_fold_disadvantage']}-fold\n"
          f"GC-Tm r={summary['GC_Tm_pearson_r']}  "
          f"very stable: {summary['pct_very_stable_nullomers']}%")


if __name__ == "__main__":
    main()