import os import numpy as np import pandas as pd from itertools import product as iproduct from Bio.Seq import Seq from scipy.stats import spearmanr, mannwhitneyu RESULTS_DIR = "results" STRESS_ELEMENTS = { "PDRE": { "consensus": "TCCGCGGA", "variants": ["TCCGCGGA", "TCCGYGGA", "TCCACGGA", "TCCGTGGA"], "factors": "Pdr1/Pdr3", }, "STRE": { "consensus": "AGGGG", "variants": ["AGGGG", "CCCCT", "AGGGA", "TCCCT"], "factors": "Msn2/Msn4", }, "HSE": { "consensus": "nGAAn", "variants": ["AGAAA", "TGAAT", "CGAAA", "GGAAT", "AGAAT", "TGAAA"], "factors": "Hsf1", }, "AP1": { "consensus": "TGASTCA", "variants": ["TGACTCA", "TGAGTCA", "TTAGTCA", "TTACTAA", "TGACTAA"], "factors": "Yap1", }, } IUPAC = { "R": ["A", "G"], "Y": ["C", "T"], "S": ["G", "C"], "W": ["A", "T"], "K": ["G", "T"], "M": ["A", "C"], "B": ["C", "G", "T"], "D": ["A", "G", "T"], "H": ["A", "C", "T"], "V": ["A", "C", "G"], "N": ["A", "C", "G", "T"], } DRUG_EFFLUX_TYPES = {"Drug efflux", "Weak acid efflux"} def expand_iupac(motif): positions = [IUPAC.get(c, [c]) for c in motif.upper()] return ["".join(combo) for combo in iproduct(*positions)] def find_motif(sequence, motif): positions = [] seq = sequence.upper() for expanded in expand_iupac(motif): ml = len(expanded) for i in range(len(seq) - ml + 1): if seq[i:i + ml] == expanded: positions.append({"position": i, "strand": "+", "sequence": expanded}) rc = str(Seq(expanded).reverse_complement()) for i in range(len(seq) - len(rc) + 1): if seq[i:i + len(rc)] == rc: positions.append({"position": i, "strand": "-", "sequence": rc}) seen = set() unique = [] for p in positions: key = (p["position"], p["strand"]) if key not in seen: unique.append(p) seen.add(key) return unique def scan_stress_elements(abc_sequences): results = {} for gene, seqs in abc_sequences.items(): results[gene] = {} for etype, einfo in STRESS_ELEMENTS.items(): all_pos = [] for motif in [einfo["consensus"]] + einfo["variants"]: all_pos.extend(find_motif(seqs["promoter"], motif)) seen = set() unique = [] for p in all_pos: key = (p["position"], p["strand"]) if key not in seen: unique.append(p) seen.add(key) results[gene][etype] = unique return results def build_correlation_df(abc_sequences, nem_results, stress_results, abc_info): rows = [] for gene in abc_sequences: info = abc_info[gene] prom_len = len(abc_sequences[gene]["promoter"]) prom_nems = len(nem_results.get(gene, {}).get("promoter", [])) nem_density = (prom_nems / prom_len) * 1000 if prom_len > 0 else 0 counts = {etype: len(stress_results[gene][etype]) for etype in STRESS_ELEMENTS} rows.append({ "gene": gene, "promoter_length": prom_len, "promoter_nems": prom_nems, "nem_density_per_kb": nem_density, "total_stress_elements": sum(counts.values()), **counts, "type": info["type"], "essential": info["essential"], "stress": info["stress"], "subfamily": info["subfamily"], "is_drug_efflux": info["type"] in DRUG_EFFLUX_TYPES, }) return pd.DataFrame(rows) def motif_disruption(abc_sequences, nem_results, stress_results): rows = [] for gene in abc_sequences: prom_nems = nem_results.get(gene, {}).get("promoter", []) for nem in prom_nems: pos = nem["position"] for etype, einfo in STRESS_ELEMENTS.items(): ml = len(einfo["consensus"]) for elem in stress_results[gene][etype]: estart = elem["position"] if estart <= pos < estart + ml: rows.append({ "gene": gene, "nem_position": pos, "nem_mutation": nem["mutation"], "element_type": etype, "motif_position": estart, "motif_strand": elem["strand"], "position_in_motif": pos - estart, }) break return pd.DataFrame(rows) def main(): from importlib.util import spec_from_file_location, module_from_spec nem_csv = os.path.join(RESULTS_DIR, "nem_comprehensive_summary.csv") if not os.path.exists(nem_csv): raise FileNotFoundError(f"{nem_csv} not found. Run 02_nem_analysis.py first.") spec = spec_from_file_location("nem_mod", "02_nem_analysis.py") mod = module_from_spec(spec) spec.loader.exec_module(mod) nullomers = mod.load_nullomers(os.path.join(RESULTS_DIR, "nullomers_k11.txt")) gene_coords = mod.parse_gff(os.path.join("data", "yeast.gff3.gz")) genome_dict = mod.load_genome_dict(os.path.join("data", "yeast_genome.fsa")) abc_info = mod.ABC_TRANSPORTERS abc_sequences = {} for gene in abc_info: if gene in gene_coords: seqs = mod.extract_sequences( gene, gene_coords, genome_dict, mod.PROMOTER_LENGTH, mod.DOWNSTREAM_LENGTH ) if seqs: abc_sequences[gene] = seqs nem_results = {} for gene in abc_sequences: nem_results[gene] = { "gene": mod.find_nems(abc_sequences[gene]["gene"], nullomers, mod.K), "promoter": mod.find_nems(abc_sequences[gene]["promoter"], nullomers, mod.K), "downstream": mod.find_nems(abc_sequences[gene]["downstream"], nullomers, mod.K), } stress_results = scan_stress_elements(abc_sequences) corr_df = build_correlation_df(abc_sequences, nem_results, stress_results, abc_info) corr_df.to_csv(os.path.join(RESULTS_DIR, "stress_element_nem_correlation.csv"), index=False) disrupt_df = motif_disruption(abc_sequences, nem_results, stress_results) if not disrupt_df.empty: disrupt_df.to_csv(os.path.join(RESULTS_DIR, "motif_disruption_by_nems.csv"), index=False) rho, p = spearmanr(corr_df["PDRE"], corr_df["nem_density_per_kb"]) drug = corr_df[corr_df["is_drug_efflux"]]["nem_density_per_kb"].values other = corr_df[~corr_df["is_drug_efflux"]]["nem_density_per_kb"].values _, p_drug = mannwhitneyu(drug, other, alternative="two-sided") print(f"PDRE-NEM rho={rho:.3f} p={p:.2e} drug efflux p={p_drug:.4f} " f"disruptions={len(disrupt_df):,}") if __name__ == "__main__": main()