Add MM-PBSA scripts + summary (exclude work dirs)

revision_exec/analysis/mmpbsa/README.md

@@ -0,0 +1,47 @@
+# MM-PBSA (`gmx_MMPBSA`)
+
+This folder contains the **dimer rep1/rep2/rep3 last-50ns (350–400 ns)** MM-PBSA-GB runs used for the revision.
+
+## Current status on this node (2026-05-06)
+
+- `gmx_MMPBSA` runs in conda env `mmpbsa` (`gmx_MMPBSA v1.5.0.3`).
+- `cpptraj` is required by `gmx_MMPBSA`. On this node it is available (may be provided via a local fix if the base env is missing it).
+
+## Numerical overflow in bond terms (known issue)
+
+Some `sander` GB outputs may show fixed-width overflow in bond terms (e.g., `BOND = *************`). Those bond energy terms were **excluded from per-frame reporting**. This does **not** affect binding free energy reporting because bond contributions cancel upon subtraction (Complex − Receptor − Ligand); the reported **ΔTOTAL** is unaffected.
+
+## Fix: create a clean MM-PBSA env that includes `cpptraj`
+
+On some images, the existing `mmpbsa` env may have AmberTools installed without `cpptraj`.
+The robust fix is to create a fresh env pinned to a supported Python version:
+
+```bash
+CONDA_SOLVER=classic conda create -n mmpbsa_py311 -c conda-forge -c bioconda \
+  python=3.11 gmx_mmpbsa ambertools mpi4py -y
+
+conda run -n mmpbsa_py311 which cpptraj
+conda run -n mmpbsa_py311 gmx_MMPBSA -h
+```
+
+## Run (GB, last 50 ns)
+
+Smoke test (few frames, sanity check):
+
+```bash
+MMPBSA_ENV=mmpbsa_py311 bash revision_exec/analysis/mmpbsa/run_dimer_rep1_last50ns_mmpbsa.sh
+```
+
+Subsampled full-window GB-only (default interval targets ~50 snapshots in 350–400 ns):
+
+```bash
+MMPBSA_ENV=mmpbsa_py311 MMPBSA_MODE=full bash revision_exec/analysis/mmpbsa/run_dimer_rep1_last50ns_mmpbsa.sh
+```
+
+Outputs are written under:
+- `revision_exec/analysis/mmpbsa/work_dimer_rep{1,2,3}_last50ns_smoke/`
+- `revision_exec/analysis/mmpbsa/work_dimer_rep{1,2,3}_last50ns_full/`
+
+Summary CSV (all three replicates):
+- `revision_exec/analysis/mmpbsa/mmpbsa_summary.csv`
+

revision_exec/analysis/mmpbsa/mmpbsa_dimer_rep1_last50ns_gb.in

@@ -0,0 +1,27 @@
+# Production-style GB run on 350–400 ns segment (md_400ns.part0005.xtc).
+# ~5000 frames at 10 ps → interval=250 ≈ 20 snapshots (edit as needed).
+
+&general
+  sys_name             = "dimer_rep1_last50ns_gb"
+  startframe           = 1
+  endframe             = 9999999
+  # 350–400 ns segment saved every 10 ps → ~5000 frames. interval=100 ≈ 50 snapshots (~1 ns spacing).
+  interval             = 100
+  forcefields          = "oldff/leaprc.ff99SB,leaprc.gaff"
+  ions_parameters      = 1
+  PBRadii              = 3
+  temperature          = 298.15
+  # IMPORTANT: leave empty so gmx_MMPBSA uses $PATH for ALL external programs
+  # (cpptraj/tleap/parmchk2/sander/gmx). If set, it must be a PATH-like string.
+  gmx_path             = ""
+  keep_files           = 2
+  solvated_trajectory  = 1
+  verbose              = 1
+/
+
+&gb
+  igb                  = 5
+  intdiel              = 1.0
+  extdiel              = 78.5
+  saltcon              = 0.0
+/

revision_exec/analysis/mmpbsa/mmpbsa_dimer_rep1_smoke_gb.in

@@ -0,0 +1,26 @@
+# Smoke test: few frames on md_400ns.part0005.xtc (350–400 ns), GB only.
+# Receptor = GROMACS index group 1 (Protein), ligand = 13 (CPP) — see revision_exec/input/index.ndx
+
+&general
+  sys_name             = "dimer_rep1_last50ns_smoke"
+  startframe           = 1
+  endframe             = 12
+  interval             = 3
+  forcefields          = "oldff/leaprc.ff99SB,leaprc.gaff"
+  ions_parameters      = 1
+  PBRadii              = 3
+  temperature          = 298.15
+  # IMPORTANT: leave empty so gmx_MMPBSA uses $PATH for ALL external programs
+  # (cpptraj/tleap/parmchk2/sander/gmx). If set, it must be a PATH-like string.
+  gmx_path             = ""
+  keep_files           = 2
+  solvated_trajectory  = 1
+  verbose              = 1
+/
+
+&gb
+  igb                  = 5
+  intdiel              = 1.0
+  extdiel              = 78.5
+  saltcon              = 0.0
+/

revision_exec/analysis/mmpbsa/mmpbsa_summary.csv

@@ -0,0 +1,7 @@
+replicate,n_frames,delta_vdw_avg_kcal_mol,delta_vdw_sd_kcal_mol,delta_eel_avg_kcal_mol,delta_eel_sd_kcal_mol,delta_egb_avg_kcal_mol,delta_egb_sd_kcal_mol,delta_esurf_avg_kcal_mol,delta_esurf_sd_kcal_mol,delta_total_avg_kcal_mol,delta_total_sd_kcal_mol
+rep1,51,-40.57,3.07,-4.95,3.52,23.01,3.26,-5.4,0.18,-27.92,3.33
+rep2,51,-42.07,1.99,-4.34,3.53,22.04,3.09,-5.57,0.14,-29.94,2.4
+rep3,51,-43.23,2.87,-8.45,28.33,21.83,25.9,-5.86,0.14,-35.71,4.58
+
+aggregate_across_reps,,mean_of_rep_avgs,sd_of_rep_avgs
+ΔTOTAL,,-31.19,4.042635279121775

revision_exec/analysis/mmpbsa/run_dimer_rep1_last50ns_mmpbsa.sh

@@ -0,0 +1,85 @@
+#!/usr/bin/env bash
+# MM-PBSA (gmx_MMPBSA) for heterodimer rep1, trajectory window 350–400 ns only.
+# Prerequisites: conda env `mmpbsa`, GROMACS in gmx-lite (path in mmpbsa input).
+#
+# Usage:
+#   bash run_dimer_rep1_last50ns_mmpbsa.sh              # smoke (few frames)
+#   MMPBSA_MODE=full bash run_dimer_rep1_last50ns_mmpbsa.sh   # subsampled full window
+#
+set -eo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+REV="$(cd "$SCRIPT_DIR/../.." && pwd)"
+MMPBSA_MODE="${MMPBSA_MODE:-smoke}"
+MMPBSA_ENV="${MMPBSA_ENV:-mmpbsa}"
+REP="${REP:-${1:-rep1}}"
+
+MMPBSA_PREFIX="/hpc2hdd/home/jyang577/miniconda3/envs/${MMPBSA_ENV}"
+GMX_MMPBSA="${MMPBSA_PREFIX}/bin/gmx_MMPBSA"
+CPPTRAJ="${MMPBSA_PREFIX}/bin/cpptraj"
+
+[[ -x "$GMX_MMPBSA" ]] || { echo "ERROR: missing executable $GMX_MMPBSA" >&2; exit 1; }
+[[ -x "$CPPTRAJ" ]] || { echo "ERROR: missing executable $CPPTRAJ" >&2; exit 1; }
+
+# Ensure external programs are discoverable via $PATH (gmx_MMPBSA uses $PATH
+# when gmx_path="" in the input file).
+export PATH="${MMPBSA_PREFIX}/bin:/hpc2hdd/home/jyang577/miniconda3/envs/gmx-lite/bin.AVX2_256:${PATH}"
+
+PROD="$REV/$REP/prod"
+NDX="$REV/input/index.ndx"
+TPR="$PROD/md_400ns.tpr"
+
+if [[ -f "$PROD/md_400ns.part0005.xtc" ]]; then
+  TRJ="$PROD/md_400ns.part0005.xtc"
+elif [[ -f "$PROD/md_400ns.part0004.xtc" ]]; then
+  TRJ="$PROD/md_400ns.part0004.xtc"
+else
+  echo "ERROR: could not find last-segment xtc under $PROD (expected md_400ns.part0004/0005.xtc)" >&2
+  exit 1
+fi
+
+# gmx_MMPBSA expects `-cp` to be a `.top`. Default to the preparation topology.
+# If you want to use a different `.top`, set TOPOLOGY_FILE=/path/to/topol.top.
+TOP_DEFAULT="$REV/prep/gate_topol.top"
+TOP="${TOPOLOGY_FILE:-$TOP_DEFAULT}"
+
+for f in "$NDX" "$TOP" "$TPR" "$TRJ"; do
+  [[ -f "$f" ]] || { echo "ERROR: missing $f" >&2; exit 1; }
+done
+
+if [[ "$MMPBSA_MODE" == "full" ]]; then
+  INFILE="$SCRIPT_DIR/mmpbsa_dimer_rep1_last50ns_gb.in"
+  OUTTAG="last50ns_gb_subsample"
+else
+  INFILE="$SCRIPT_DIR/mmpbsa_dimer_rep1_smoke_gb.in"
+  OUTTAG="smoke_gb"
+fi
+
+[[ -f "$INFILE" ]] || { echo "ERROR: missing input $INFILE" >&2; exit 1; }
+
+WORKDIR="$SCRIPT_DIR/work_dimer_${REP}_last50ns_${MMPBSA_MODE}"
+mkdir -p "$WORKDIR"
+cd "$WORKDIR"
+
+LOG="$WORKDIR/mmpbsa_${OUTTAG}_$(date +%Y%m%d_%H%M%S).log"
+
+echo "WORKDIR=$WORKDIR" | tee "$LOG"
+echo "INFILE=$INFILE" | tee -a "$LOG"
+echo "TPR=$TPR" | tee -a "$LOG"
+echo "TRJ=$TRJ" | tee -a "$LOG"
+
+# Protein = index 1, CPP = 13 (0-based group order in index.ndx)
+set +e
+"$GMX_MMPBSA" -O -nogui \
+  -i "$INFILE" \
+  -cs "$TPR" \
+  -ci "$NDX" \
+  -cg 1 13 \
+  -cp "$TOP" \
+  -ct "$TRJ" \
+  -eo "${OUTTAG}_perframe.csv" \
+  -o "${OUTTAG}_FINAL_RESULTS.dat" \
+  2>&1 | tee -a "$LOG"
+rc=${PIPESTATUS[0]}
+set -e
+exit "$rc"

revision_exec/analysis/mmpbsa/summarize_mmpbsa_gb_last50ns.py

@@ -0,0 +1,128 @@
+#!/usr/bin/env python3
+from __future__ import annotations
+
+import argparse
+import csv
+import math
+from dataclasses import dataclass
+from pathlib import Path
+from statistics import mean, stdev
+
+
+@dataclass(frozen=True)
+class DeltaStats:
+    avg: float
+    sd: float
+
+
+DELTA_KEYS = ("ΔVDWAALS", "ΔEEL", "ΔEGB", "ΔESURF", "ΔTOTAL")
+
+
+def _parse_float(s: str) -> float:
+    s = s.strip()
+    if s.lower() == "nan":
+        return float("nan")
+    return float(s)
+
+
+def parse_final_results(path: Path) -> dict[str, DeltaStats]:
+    """Parse gmx_MMPBSA FINAL_RESULTS file and return delta avg + SD (SD column, not SD(Prop.))."""
+    txt = path.read_text(errors="replace").splitlines()
+    in_delta = False
+    out: dict[str, DeltaStats] = {}
+
+    for line in txt:
+        if line.startswith("Delta (Complex - Receptor - Ligand):"):
+            in_delta = True
+            continue
+        if not in_delta:
+            continue
+        if not line.strip():
+            continue
+        if line.startswith("Δ"):
+            parts = line.split()
+            if len(parts) < 4:
+                continue
+            key = parts[0]
+            if key not in DELTA_KEYS:
+                # Skip terms like "Δ1-4 VDW" whose key is split across tokens.
+                continue
+            avg = _parse_float(parts[1])
+            sd = _parse_float(parts[3])
+            out[key] = DeltaStats(avg=avg, sd=sd)
+
+    missing = [k for k in DELTA_KEYS if k not in out]
+    if missing:
+        raise ValueError(f"Missing delta terms {missing} in {path}")
+    return out
+
+
+def safe_stdev(xs: list[float]) -> float:
+    xs2 = [x for x in xs if not math.isnan(x)]
+    if len(xs2) < 2:
+        return float("nan")
+    return float(stdev(xs2))
+
+
+def main() -> None:
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--rep1", required=True, type=Path)
+    ap.add_argument("--rep2", required=True, type=Path)
+    ap.add_argument("--rep3", required=True, type=Path)
+    ap.add_argument("--out-csv", required=True, type=Path)
+    ap.add_argument("--n-frames", type=int, default=51)
+    args = ap.parse_args()
+
+    reps = {
+        "rep1": parse_final_results(args.rep1),
+        "rep2": parse_final_results(args.rep2),
+        "rep3": parse_final_results(args.rep3),
+    }
+
+    args.out_csv.parent.mkdir(parents=True, exist_ok=True)
+    with args.out_csv.open("w", newline="") as f:
+        w = csv.writer(f)
+        w.writerow(
+            [
+                "replicate",
+                "n_frames",
+                "delta_vdw_avg_kcal_mol",
+                "delta_vdw_sd_kcal_mol",
+                "delta_eel_avg_kcal_mol",
+                "delta_eel_sd_kcal_mol",
+                "delta_egb_avg_kcal_mol",
+                "delta_egb_sd_kcal_mol",
+                "delta_esurf_avg_kcal_mol",
+                "delta_esurf_sd_kcal_mol",
+                "delta_total_avg_kcal_mol",
+                "delta_total_sd_kcal_mol",
+            ]
+        )
+
+        for rep, d in reps.items():
+            w.writerow(
+                [
+                    rep,
+                    args.n_frames,
+                    d["ΔVDWAALS"].avg,
+                    d["ΔVDWAALS"].sd,
+                    d["ΔEEL"].avg,
+                    d["ΔEEL"].sd,
+                    d["ΔEGB"].avg,
+                    d["ΔEGB"].sd,
+                    d["ΔESURF"].avg,
+                    d["ΔESURF"].sd,
+                    d["ΔTOTAL"].avg,
+                    d["ΔTOTAL"].sd,
+                ]
+            )
+
+        tot_avgs = [reps[r]["ΔTOTAL"].avg for r in ("rep1", "rep2", "rep3")]
+        w.writerow([])
+        w.writerow(["aggregate_across_reps", "", "mean_of_rep_avgs", "sd_of_rep_avgs"])
+        w.writerow(["ΔTOTAL", "", float(mean(tot_avgs)), safe_stdev(tot_avgs)])
+
+
+if __name__ == "__main__":
+    main()
+