py2ls 0.1.10.12__py3-none-any.whl → 0.2.7.10__py3-none-any.whl

py2ls/rna2ls.py

@@ -0,0 +1,311 @@
+#!/usr/bin/env python3
+"""
+rna_master_tools.py
+Single-file advanced RNA-seq toolkit + CLI.
+
+Usage examples:
+    python rna_master_tools.py run-rmats --b1 a1.bam,a2.bam --b2 b1.bam,b2.bam --gtf ref.gtf --outdir rmats_out
+    python rna_master_tools.py parse-rmats --in rmats_out/SE.MATS.JCEC.txt --out rmats_parsed.tsv
+    python rna_master_tools.py run-starfusion --left left.fq --right right.fq --genome_lib_dir GRCh38_ctat_lib --outdir fusion_out
+    python rna_master_tools.py compute-wgcna --counts counts_normalized.tsv --out modules.tsv
+    python rna_master_tools.py biomarkers --counts counts_normalized.tsv --labels samplesheet.tsv --out biomarker_report.tsv
+"""
+
+import os
+import sys
+import argparse
+import logging
+import subprocess
+import numpy as np
+import pandas as pd
+import matplotlib.pyplot as plt
+
+from sklearn.ensemble import RandomForestClassifier
+from sklearn.model_selection import train_test_split
+from sklearn.cluster import AgglomerativeClustering
+from sklearn.metrics import classification_report
+
+logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s] %(message)s")
+logger = logging.getLogger(__name__)
+
+# -------------------------
+# Utilities
+# -------------------------
+def ensure_dir(path):
+    os.makedirs(path, exist_ok=True)
+    return path
+
+def run_cmd(cmd, check=True):
+    logger.info("Run cmd: %s", " ".join(cmd))
+    try:
+        res = subprocess.run(cmd, check=check, capture_output=True, text=True)
+        logger.debug("stdout: %s", res.stdout[:200])
+        logger.debug("stderr: %s", res.stderr[:200])
+        return res
+    except subprocess.CalledProcessError as e:
+        logger.error("Command failed: %s", e.stderr)
+        raise
+
+def read_table(path, index_col=None):
+    return pd.read_csv(path, sep="\t", index_col=index_col)
+
+def write_table(df, path):
+    df.to_csv(path, sep="\t", index=True)
+
+# -------------------------
+# rMATS wrappers & parsers
+# -------------------------
+def run_rmats_cli(b1_list, b2_list, gtf, outdir, readlen=100, threads=8, rmats_cmd="rmats.py"):
+    """
+    Run rMATS CLI. b1_list/b2_list are comma-separated BAM paths strings.
+    """
+    ensure_dir(outdir)
+    tmpdir = ensure_dir(os.path.join(outdir, "tmp"))
+    cmd = [
+        rmats_cmd,
+        "--b1", b1_list,
+        "--b2", b2_list,
+        "--gtf", gtf,
+        "--od", outdir,
+        "--tmp", tmpdir,
+        "--readLength", str(readlen),
+        "--nthread", str(threads)
+    ]
+    run_cmd(cmd)
+    return outdir
+
+def parse_rmats_event_table(event_file, out_tsv=None):
+    """
+    Parse rMATS event file and compute PSI means and dPSI.
+    """
+    df = pd.read_csv(event_file, sep="\t", low_memory=False)
+    # helper
+    def mean_psi(s):
+        try:
+            vals = [float(x) for x in str(s).split(",") if x not in (".", "", "NA")]
+            return np.nanmean(vals) if len(vals)>0 else np.nan
+        except:
+            return np.nan
+    if 'IncLevel1' in df.columns:
+        df['PSI_group1'] = df['IncLevel1'].apply(mean_psi)
+    if 'IncLevel2' in df.columns:
+        df['PSI_group2'] = df['IncLevel2'].apply(mean_psi)
+    if 'PSI_group1' in df and 'PSI_group2' in df:
+        df['dPSI'] = df['PSI_group1'] - df['PSI_group2']
+    for col in ['PValue','FDR']:
+        if col in df.columns:
+            df[col] = pd.to_numeric(df[col], errors='coerce')
+    if out_tsv:
+        df.to_csv(out_tsv, sep="\t", index=False)
+    return df
+
+def plot_rmats_volcano(df, out_png=None, fdr_thresh=0.05, dpsi_thresh=0.1):
+    neglog = -np.log10(df['FDR'].replace(0,1e-300))
+    sig = (df['FDR'] <= fdr_thresh) & (df['dPSI'].abs() >= dpsi_thresh)
+    plt.figure(figsize=(7,6))
+    plt.scatter(df['dPSI'], neglog, s=8, c='gray', alpha=0.7)
+    if sig.any():
+        plt.scatter(df.loc[sig,'dPSI'], neglog.loc[sig], s=10, c='red', alpha=0.8)
+    plt.axvline(dpsi_thresh, color='blue', linestyle='--')
+    plt.axvline(-dpsi_thresh, color='blue', linestyle='--')
+    plt.axhline(-np.log10(fdr_thresh), color='green', linestyle='--')
+    plt.xlabel('dPSI'); plt.ylabel('-log10(FDR)'); plt.title('rMATS volcano')
+    if out_png:
+        plt.savefig(out_png, dpi=150)
+    return plt
+
+# -------------------------
+# STAR-Fusion wrapper & parser
+# -------------------------
+def run_star_fusion_cli(left_fq, right_fq, genome_lib_dir, outdir, threads=8):
+    """
+    Run STAR-Fusion. Assumes STAR-Fusion installed and in PATH or in container.
+    """
+    ensure_dir(outdir)
+    cmd = [
+        "STAR-Fusion",
+        "--left_fq", left_fq,
+        "--right_fq", right_fq,
+        "--genome_lib_dir", genome_lib_dir,
+        "--output_dir", outdir,
+        "--CPU", str(threads)
+    ]
+    run_cmd(cmd)
+    return outdir
+
+def parse_star_fusion_tsv(fusion_tsv, out_tsv=None):
+    df = pd.read_csv(fusion_tsv, sep="\t", comment='#', low_memory=False)
+    if out_tsv:
+        df.to_csv(out_tsv, sep="\t", index=False)
+    return df
+
+# -------------------------
+# WGCNA-like coexpression (lightweight)
+# -------------------------
+def compute_wgcna_modules_cli(counts_tsv, out_tsv=None, soft_power=6):
+    """
+    counts_tsv: genes x samples (rows genes, columns samples) or samples x genes.
+    We'll accept genes rows (index=gene).
+    Returns DataFrame gene->module
+    """
+    df = pd.read_csv(counts_tsv, sep="\t", index_col=0)
+    # ensure genes are rows
+    if df.shape[0] < df.shape[1] and df.index.str.contains('ENS').any():
+        # likely already genes x samples; keep
+        pass
+    # compute correlation between genes (use transpose if needed)
+    expr = df
+    # if many samples < genes, correlation across samples => gene-gene correlation requires samples dimension
+    corr = expr.T.corr()  # gene x gene
+    adj = np.abs(corr) ** soft_power
+    # approximate TOM (simple)
+    tom = adj * np.abs(corr)
+    dist = 1 - tom
+    # fill nans
+    dist = dist.fillna(1.0)
+    # clustering
+    # AgglomerativeClustering does not accept distance matrix directly for fit_predict unless precomputed affinity
+    clustering = AgglomerativeClustering(n_clusters=None, distance_threshold=0.2, affinity='precomputed', linkage='average')
+    labels = clustering.fit_predict(dist.values)
+    modules = pd.DataFrame({'gene': corr.index, 'module': labels})
+    if out_tsv:
+        modules.to_csv(out_tsv, sep="\t", index=False)
+    return modules
+
+# -------------------------
+# BIOMARKER SELECTION + ML
+# -------------------------
+def select_features_rf_cli(counts_tsv, labels_tsv, out_tsv=None, topk=50):
+    """
+    counts_tsv: genes x samples (index gene, columns samples) OR samples x genes.
+    labels_tsv: must include columns sampleID, label
+    We'll convert to samples x features for ML.
+    """
+    counts = pd.read_csv(counts_tsv, sep="\t", index_col=0)
+    ss = pd.read_csv(labels_tsv, sep="\t")
+    # convert to samples x genes if needed
+    if counts.index[0].startswith('ENS') or counts.index[0].isalpha():
+        # genes x samples -> transpose
+        X = counts.T
+    else:
+        X = counts
+    # align samples
+    if 'sampleID' in ss.columns:
+        ss = ss.set_index('sampleID')
+    labels = ss.loc[X.index, ss.columns[-1]]  # last column is label
+    # RF
+    rf = RandomForestClassifier(n_estimators=500, random_state=42, n_jobs=1)
+    rf.fit(X, labels)
+    imp = pd.Series(rf.feature_importances_, index=X.columns).sort_values(ascending=False)
+    top = imp.head(topk)
+    if out_tsv:
+        top.to_csv(out_tsv, sep="\t", header=['importance'])
+    return top
+
+def train_classifiers_cli(counts_tsv, labels_tsv, out_tsv=None, test_size=0.25):
+    counts = pd.read_csv(counts_tsv, sep="\t", index_col=0)
+    ss = pd.read_csv(labels_tsv, sep="\t")
+    if counts.index[0].startswith('ENS') or counts.index[0].isalpha():
+        X = counts.T
+    else:
+        X = counts
+    if 'sampleID' in ss.columns:
+        ss = ss.set_index('sampleID')
+    y = ss.loc[X.index, ss.columns[-1]]
+    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=test_size, random_state=42)
+    rf = RandomForestClassifier(n_estimators=400, random_state=42)
+    rf.fit(X_train, y_train)
+    pred = rf.predict(X_test)
+    rep = classification_report(y_test, pred, output_dict=True)
+    if out_tsv:
+        pd.DataFrame(rep).to_csv(out_tsv, sep="\t")
+    return rep
+
+# -------------------------
+# CLI: argparse dispatcher
+# -------------------------
+def build_parser():
+    p = argparse.ArgumentParser(prog="rna_master_tools.py")
+    sub = p.add_subparsers(dest="cmd")
+
+    # rMATS run
+    a = sub.add_parser("run-rmats")
+    a.add_argument("--b1", required=True, help="Comma-separated BAMs for group1")
+    a.add_argument("--b2", required=True, help="Comma-separated BAMs for group2")
+    a.add_argument("--gtf", required=True)
+    a.add_argument("--outdir", required=True)
+    a.add_argument("--readlen", type=int, default=100)
+    a.add_argument("--threads", type=int, default=8)
+    a.add_argument("--rmats-cmd", default="rmats.py")
+
+    # parse rMATS
+    b = sub.add_parser("parse-rmats")
+    b.add_argument("--in", dest="infile", required=True)
+    b.add_argument("--out", dest="out", default=None)
+
+    # plot rMATS volcano
+    c = sub.add_parser("plot-rmats")
+    c.add_argument("--in", dest="infile", required=True)
+    c.add_argument("--out", dest="out", default=None)
+
+    # STAR-Fusion run
+    d = sub.add_parser("run-starfusion")
+    d.add_argument("--left", required=True)
+    d.add_argument("--right", required=True)
+    d.add_argument("--genome_lib_dir", required=True)
+    d.add_argument("--outdir", required=True)
+    d.add_argument("--threads", type=int, default=8)
+
+    # parse star-fusion
+    e = sub.add_parser("parse-starfusion")
+    e.add_argument("--in", dest="infile", required=True)
+    e.add_argument("--out", dest="out", default=None)
+
+    # WGCNA
+    f = sub.add_parser("compute-wgcna")
+    f.add_argument("--counts", required=True)
+    f.add_argument("--out", dest="out", default=None)
+    f.add_argument("--power", type=int, default=6)
+
+    # biomarkers select
+    g = sub.add_parser("biomarkers")
+    g.add_argument("--counts", required=True)
+    g.add_argument("--labels", required=True)
+    g.add_argument("--out", dest="out", default=None)
+    g.add_argument("--topk", type=int, default=50)
+
+    # classifiers
+    h = sub.add_parser("train-classifiers")
+    h.add_argument("--counts", required=True)
+    h.add_argument("--labels", required=True)
+    h.add_argument("--out", dest="out", default=None)
+
+    return p
+
+def main(argv=None):
+    p = build_parser()
+    args = p.parse_args(argv)
+
+    if args.cmd == "run-rmats":
+        run_rmats_cli(args.b1, args.b2, args.gtf, args.outdir, readlen=args.readlen, threads=args.threads, rmats_cmd=args.rmats_cmd)
+    elif args.cmd == "parse-rmats":
+        parse_rmats_event_table(args.infile, out_tsv=args.out)
+    elif args.cmd == "plot-rmats":
+        df = parse_rmats_event_table(args.infile)
+        plot_rmats_volcano(df, out_png=args.out)
+    elif args.cmd == "run-starfusion":
+        run_star_fusion_cli(args.left, args.right, args.genome_lib_dir, args.outdir, threads=args.threads)
+    elif args.cmd == "parse-starfusion":
+        parse_star_fusion_tsv(args.infile, out_tsv=args.out)
+    elif args.cmd == "compute-wgcna":
+        compute_wgcna_modules_cli(args.counts, out_tsv=args.out, soft_power=args.power)
+    elif args.cmd == "biomarkers":
+        select_features_rf_cli(args.counts, args.labels, out_tsv=args.out, topk=args.topk)
+    elif args.cmd == "train-classifiers":
+        train_classifiers_cli(args.counts, args.labels, out_tsv=args.out)
+    else:
+        p.print_help()
+        sys.exit(1)
+
+if __name__ == "__main__":
+    main()