agentleFS
dailycafidailycafi/metabolism-skills/skills/metabolomics-analysis/pharmacometabolomics/SKILL.md

bio-metabolomics-analysis-pharmacometabolomics

Skill7 starsChanged 8 months agoMITView on GitHub

Study drug metabolism and pharmacokinetic responses using pharmacometabolomics workflows with RDKit, DrugBank, and MetaboAnalystR. Use when: user asks about drug metabolites, CYP450 metabolism, ADME predictions, phase I/II biotransformation, or drug-response metabolite profiling. Triggers: drug metabolism, CYP450, ADME, drug metabolite, phase I metabolism, phase II conjugation, pharmacokinetics, DrugBank, metabolic soft spots, glucuronidation, sulfation, drug-metabolite interaction, pharmacometabolomics, xenobiotic metabolism.

  • Reads credentials
  • Installs packages

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 7 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before using code patterns, verify installed versions match.

    line 12
  2. Distinctive:

    If code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API …

    line 16
  3. Distinctive:

    Approach: Use the DrugBank API to retrieve metabolizing enzymes, transporters, and known metabolites for a drug.

    line 281
  4. Distinctive:

    Approach: Use MetaboAnalystR to normalize, perform statistical testing, run pathway enrichment, and generate report-quality figures.

    line 325
  5. Distinctive:

    Always collect pre-dose baseline samples for paired comparisons

    line 396
  6. Distinctive:

    Use at least 3 time points post-dose to capture metabolic dynamics

    line 397
  7. Distinctive:

    Use internal standards for each metabolite class to correct for matrix effects

    line 401

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Version Compatibility

Reference examples tested with: RDKit 2023.09+, MetaboAnalystR 4.0+, pandas 2.0+

Before using code patterns, verify installed versions match. If versions differ:

  • Python: pip show <package> then help(module.function) to check signatures
  • R: packageVersion("<pkg>") then ?function_name to verify parameters

If code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying.

Pharmacometabolomics

"Profile drug response via metabolomics and predict drug metabolism" -> Compare pre-dose vs post-dose metabolite profiles, predict CYP450 enzyme interactions and ADME properties from molecular structure, identify phase I/II metabolites in LC-MS data, and integrate PK/PD parameters with metabolomics measurements.

  • Python: rdkit for molecular descriptors, pandas/scipy for statistics
  • R: MetaboAnalystR for differential metabolite analysis

Installation

pip install rdkit pandas numpy scipy scikit-learn matplotlib requests

# R package (for MetaboAnalystR section) — NOT on CRAN, install from GitHub
Rscript -e 'devtools::install_github("xia-lab/MetaboAnalystR", build_vignettes = FALSE)'

Drug Metabolite Profiling: Pre-dose vs Post-dose

Goal: Identify metabolites that change significantly after drug administration.

Approach: Perform paired statistical testing between pre-dose and post-dose metabolite levels, apply multiple testing correction, and generate volcano plots.

import pandas as pd
import numpy as np
from scipy import stats
from statsmodels.stats.multitest import multipletests
import matplotlib.pyplot as plt

def drug_response_analysis(pre_dose_file, post_dose_file, alpha=0.05):
    """Compare metabolite profiles before and after drug treatment."""
    pre = pd.read_csv(pre_dose_file, index_col=0)
    post = pd.read_csv(post_dose_file, index_col=0)

    common = pre.index.intersection(post.index)
    pre, post = pre.loc[common], post.loc[common]

    results = []
    for metabolite in pre.columns:
        pre_vals, post_vals = pre[metabolite].values, post[metabolite].values
        t_stat, p_val = stats.ttest_rel(post_vals, pre_vals)
        log2fc = np.log2((post_vals.mean() + 1e-10) / (pre_vals.mean() + 1e-10))
        results.append({"metabolite": metabolite, "log2_fold_change": log2fc,
                        "t_statistic": t_stat, "p_value": p_val})

    df = pd.DataFrame(results)
    _, fdr, _, _ = multipletests(df["p_value"], method="fdr_bh")
    df["fdr"] = fdr
    df["significant"] = (df["fdr"] < alpha) & (np.abs(df["log2_fold_change"]) > 0.5)
    return df.sort_values("p_value")


results = drug_response_analysis("pre_dose.csv", "post_dose.csv")
print(f"Significant changes: {results['significant'].sum()}")

# Volcano plot
fig, ax = plt.subplots(figsize=(8, 6))
colors = ["#b2182b" if s else "#cccccc" for s in results["significant"]]
ax.scatter(results["log2_fold_change"], -np.log10(results["p_value"]), c=colors, s=20)
ax.set_xlabel("Log2 Fold Change (Post/Pre)")
ax.set_ylabel("-Log10(p-value)")
plt.savefig("volcano_drug_response.png", dpi=300, bbox_inches="tight")
plt.close()

CYP450 Substrate Prediction

Goal: Predict which CYP450 enzymes are likely to metabolize a given drug.

Approach: Compute RDKit molecular descriptors relevant to CYP450 binding (lipophilicity, molecular weight, H-bond donors/acceptors, aromatic rings), then apply rule-based filtering for major CYP isoforms.

from rdkit import Chem
from rdkit.Chem import Descriptors, rdMolDescriptors

def compute_cyp450_descriptors(smiles):
    """Compute molecular descriptors relevant to CYP450 metabolism."""
    mol = Chem.MolFromSmiles(smiles)
    if mol is None:
        raise ValueError(f"Invalid SMILES: {smiles}")
    return {
        "MW": Descriptors.MolWt(mol), "LogP": Descriptors.MolLogP(mol),
        "HBA": Descriptors.NumHAcceptors(mol), "TPSA": Descriptors.TPSA(mol),
        "AromaticRings": rdMolDescriptors.CalcNumAromaticRings(mol),
    }


def predict_cyp_isoforms(descriptors):
    """Rule-based CYP450 isoform prediction based on substrate profiles."""
    predictions = []
    mw, logp, hba = descriptors["MW"], descriptors["LogP"], descriptors["HBA"]
    aromatic = descriptors["AromaticRings"]

    if mw > 300 and logp > 1.0:  # CYP3A4: larger lipophilic molecules
        predictions.append(("CYP3A4", round(min(1.0, (mw - 300) / 500 * 0.5 + logp / 6 * 0.5), 2)))
    if 200 < mw < 500 and 1.0 < logp < 4.0:  # CYP2D6: basic amines
        predictions.append(("CYP2D6", round(0.6 if aromatic >= 1 else 0.3, 2)))
    if 200 < mw < 500 and hba >= 2:  # CYP2C9: acidic compounds
        predictions.append(("CYP2C9", round(0.5 if logp > 2.0 else 0.3, 2)))
    if 200 < mw < 450 and 1.5 < logp < 5.0:  # CYP2C19
        predictions.append(("CYP2C19", 0.4))
    if aromatic >= 2 and mw < 400:  # CYP1A2: planar aromatics
        predictions.append(("CYP1A2", round(0.5 + aromatic * 0.1, 2)))

    return sorted(predictions, key=lambda x: x[1], reverse=True)


# Example: Ibuprofen
desc = compute_cyp450_descriptors("CC(C)Cc1ccc(cc1)C(C)C(=O)O")
for isoform, score in predict_cyp_isoforms(desc):
    print(f"  {isoform}: {score}")

ADME Property Prediction

Goal: Estimate absorption, distribution, metabolism, and excretion properties from molecular structure.

Approach: Compute Lipinski and Veber rule descriptors, predict Caco-2 permeability class, and estimate aqueous solubility from LogP.

from rdkit import Chem
from rdkit.Chem import Descriptors
import numpy as np

def predict_adme(smiles):
    """Predict ADME properties from molecular structure."""
    mol = Chem.MolFromSmiles(smiles)
    if mol is None:
        raise ValueError(f"Invalid SMILES: {smiles}")

    mw = Descriptors.MolWt(mol)
    logp = Descriptors.MolLogP(mol)
    hbd = Descriptors.NumHDonors(mol)
    hba = Descriptors.NumHAcceptors(mol)
    tpsa = Descriptors.TPSA(mol)
    rotbonds = Descriptors.NumRotatableBonds(mol)

    lipinski_violations = sum([mw > 500, logp > 5, hbd > 5, hba > 10])
    veber_pass = tpsa <= 140 and rotbonds <= 10

    # ESOL solubility model (Delaney 2004)
    aromatic_atoms = sum(1 for atom in mol.GetAtoms() if atom.GetIsAromatic())
    arom_prop = aromatic_atoms / max(mol.GetNumHeavyAtoms(), 1)
    log_s = 0.16 - 0.63 * logp - 0.0062 * mw + 0.066 * rotbonds - 0.74 * arom_prop

    sol_class = ("Highly soluble" if log_s > -2 else "Soluble" if log_s > -4
                 else "Moderately soluble" if log_s > -6 else "Poorly soluble")

    return {
        "MW": round(mw, 1), "LogP": round(logp, 2),
        "HBD": hbd, "HBA": hba, "TPSA": round(tpsa, 1),
        "Lipinski_pass": lipinski_violations <= 1, "Veber_pass": veber_pass,
        "LogS_ESOL": round(log_s, 2), "Solubility": sol_class,
        "GI_absorption": "High" if tpsa < 140 else "Low",
        "BBB_permeant": "Yes" if tpsa < 80 and logp > 0 else "No",
    }


# Example: Metformin
for prop, val in predict_adme("CN(C)C(=N)NC(=N)N").items():
    print(f"  {prop}: {val}")

Metabolic Soft Spot Prediction

Goal: Identify likely sites of metabolic oxidation on a drug molecule.

Approach: Compute atomic partial charges and identify electron-rich centers (aromatic carbons, allylic positions, benzylic carbons) that are susceptible to CYP-mediated oxidation.

from rdkit import Chem
from rdkit.Chem import AllChem, Descriptors

def predict_soft_spots(smiles):
    """Predict metabolic soft spots (likely sites of CYP oxidation)."""
    mol = Chem.MolFromSmiles(smiles)
    if mol is None:
        raise ValueError(f"Invalid SMILES: {smiles}")

    mol = Chem.AddHs(mol)
    AllChem.ComputeGasteigerCharges(mol)
    soft_spots = []

    for atom in mol.GetAtoms():
        if atom.GetSymbol() == "H":
            continue
        idx, symbol = atom.GetIdx(), atom.GetSymbol()
        reasons = []

        if atom.GetIsAromatic() and symbol == "C":
            reasons.append("aromatic_hydroxylation")
        if symbol == "C" and not atom.GetIsAromatic():
            if any(n.GetIsAromatic() for n in atom.GetNeighbors()):
                reasons.append("benzylic_oxidation")
        if symbol == "N" and atom.GetDegree() == 3 and not atom.GetIsAromatic():
            reasons.append("N_dealkylation")
        if symbol == "O" and atom.GetDegree() == 2:
            if "C" in [n.GetSymbol() for n in atom.GetNeighbors()]:
                reasons.append("O_dealkylation")
        if symbol == "S" and atom.GetDegree() <= 2:
            reasons.append("S_oxidation")

        if reasons:
            soft_spots.append((idx, symbol, reasons))

    return soft_spots


# Example: Verapamil
for idx, symbol, reasons in predict_soft_spots("COc1ccc(CCN(C)CCCC(C#N)(c2ccc(OC)c(OC)c2)C(C)C)cc1OC"):
    print(f"  Atom {idx} ({symbol}): {', '.join(reasons)}")

Phase I/II Metabolite Identification from LC-MS

Goal: Identify expected phase I and phase II metabolites in LC-MS data by computing theoretical m/z shifts.

Approach: Define mass shifts for common biotransformations, then search extracted ion chromatograms for matching peaks.

import pandas as pd
import numpy as np

# Common phase I/II biotransformation mass shifts (Da)
BIOTRANSFORMATIONS = {
    "Hydroxylation (+O)": 15.9949, "Dihydroxylation (+2O)": 31.9898,
    "Dehydrogenation (-2H)": -2.0157, "N-dealkylation (-CH2)": -14.0157,
    "Reduction (+2H)": 2.0157,
    "Glucuronidation (+C6H8O6)": 176.0321, "Sulfation (+SO3)": 79.9568,
    "Glutathione (+GSH)": 305.0682, "Acetylation (+C2H2O)": 42.0106,
    "Methylation (+CH2)": 14.0157, "Glycine conjugation": 57.0215,
}


def search_metabolites(parent_mz, ms_data_file, ppm_tolerance=10):
    """Search LC-MS data for predicted drug metabolites."""
    ms_data = pd.read_csv(ms_data_file)
    candidates = []
    for name, shift in BIOTRANSFORMATIONS.items():
        expected = parent_mz + shift
        tol = expected * ppm_tolerance / 1e6
        matches = ms_data[(ms_data["mz"] >= expected - tol) &
                          (ms_data["mz"] <= expected + tol)]
        if len(matches) > 0:
            best = matches.loc[matches["intensity"].idxmax()]
            candidates.append({
                "transformation": name, "expected_mz": round(expected, 4),
                "observed_mz": round(best["mz"], 4),
                "ppm_error": round((best["mz"] - expected) / expected * 1e6, 1),
                "retention_time": round(best["rt"], 2), "intensity": best["intensity"],
            })
    return pd.DataFrame(candidates).sort_values("intensity", ascending=False)


# Example: Search for metabolites of verapamil ([M+H]+ = 455.2904)
metabolites = search_metabolites(455.2904, "lcms_postdose.csv", ppm_tolerance=5)
print(metabolites.to_string(index=False))

DrugBank API Integration

Goal: Query DrugBank for known drug-metabolite interactions and enzyme associations.

Approach: Use the DrugBank API to retrieve metabolizing enzymes, transporters, and known metabolites for a drug.

import os
import requests

# DrugBank API requires a paid subscription (https://go.drugbank.com/api)
DRUGBANK_API_KEY = os.environ.get("DRUGBANK_API_KEY")
if not DRUGBANK_API_KEY:
    raise EnvironmentError("Set DRUGBANK_API_KEY environment variable (requires paid subscription)")

def query_drugbank(drugbank_id):
    """Query DrugBank API for drug metabolism information."""
    headers = {"Authorization": f"Bearer {DRUGBANK_API_KEY}"}
    base_url = "https://api.drugbank.com/v1"

    resp = requests.get(f"{base_url}/drugs/{drugbank_id}", headers=headers)
    resp.raise_for_status()
    drug = resp.json()

    return {
        "drug_name": drug.get("name"),
        "metabolizing_enzymes": [
            {"name": e.get("name"), "gene": e.get("gene_name")}
            for e in drug.get("enzymes", [])
        ],
        "transporters": [
            {"name": t.get("name"), "gene": t.get("gene_name")}
            for t in drug.get("transporters", [])
        ],
        "half_life": drug.get("half_life"),
    }


info = query_drugbank("DB00661")  # Verapamil
print(f"Drug: {info['drug_name']}, Half-life: {info['half_life']}")
for enz in info["metabolizing_enzymes"]:
    print(f"  Enzyme: {enz['name']} ({enz['gene']})")

MetaboAnalystR: Differential Drug Response Analysis

Goal: Perform comprehensive differential analysis of metabolomics data comparing drug responders vs non-responders.

Approach: Use MetaboAnalystR to normalize, perform statistical testing, run pathway enrichment, and generate report-quality figures.

library(MetaboAnalystR)

# Initialize and load data (rows = samples, cols = metabolites, first col = group)
mSet <- InitDataObjects("conc", "stat", FALSE)
mSet <- Read.TextData(mSet, "drug_response_metabolomics.csv", "rowu", "disc")
mSet <- SanityCheckData(mSet)
mSet <- ReplaceMin(mSet)

# Normalize: log transform + auto-scaling
mSet <- PreparePrenormData(mSet)
mSet <- Normalization(mSet, "NULL", "LogNorm", "AutoNorm", ref = NULL,
                       ratio = FALSE, ratioNum = 20)

# Volcano plot (fold change + t-test combined)
mSet <- FC.Anal(mSet, fc.thresh = 1.5, cmpType = 0)
mSet <- Ttests.Anal(mSet, nonpar = FALSE, threshp = 0.05,
                     paired = FALSE, equal.var = TRUE)
mSet <- Volcano.Anal(mSet, FALSE, 1.5, 0, 0.1, TRUE, 0.05, TRUE, "raw")
mSet <- PlotVolcano(mSet, "volcano", format = "png", dpi = 300, width = NA)

# PLS-DA: classify responders vs non-responders
mSet <- PLSR.Anal(mSet, reg = TRUE)
mSet <- PlotPLS2DScore(mSet, "plsda_score", format = "png", dpi = 300,
                        width = NA, 1, 2, 0.95, 0)
mSet <- PlotPLS.Imp(mSet, "plsda_vip", format = "png", dpi = 300,
                     width = NA, "vip", "Comp. 1", 20)

PK/PD Integration with Metabolomics

Goal: Correlate PK parameters (AUC, Cmax, Tmax) with metabolomics profiles to identify metabolic signatures of drug exposure.

import pandas as pd
from scipy.stats import spearmanr
from statsmodels.stats.multitest import multipletests

def pk_metabolomics_correlation(pk_file, metab_file):
    """Correlate PK parameters (AUC, Cmax, Tmax) with metabolite levels."""
    pk = pd.read_csv(pk_file, index_col="subject_id")
    metab = pd.read_csv(metab_file, index_col="subject_id")
    common = pk.index.intersection(metab.index)
    pk, metab = pk.loc[common], metab.loc[common]

    results = []
    for pk_param in pk.columns:
        for met_name in metab.columns:
            rho, pval = spearmanr(pk[pk_param], metab[met_name])
            results.append({"pk_parameter": pk_param, "metabolite": met_name,
                            "spearman_rho": round(rho, 3), "p_value": pval})

    df = pd.DataFrame(results)
    corrected = []
    for param in pk.columns:
        subset = df[df["pk_parameter"] == param].copy()
        _, fdr, _, _ = multipletests(subset["p_value"], method="fdr_bh")
        subset["fdr"] = fdr
        corrected.append(subset)
    return pd.concat(corrected).sort_values("p_value")


correlations = pk_metabolomics_correlation("pk_parameters.csv", "metabolomics.csv")
sig_corr = correlations[correlations["fdr"] < 0.05]
print(f"Significant PK-metabolite correlations: {len(sig_corr)}")

Best Practices

  • Always collect pre-dose baseline samples for paired comparisons
  • Use at least 3 time points post-dose to capture metabolic dynamics
  • Include BMI, age, sex, and CYP genotype as covariates
  • Set mass tolerance based on instrument: TOF 10-20 ppm, Orbitrap 3-5 ppm
  • Confirm putative metabolites with MS/MS fragmentation matching to parent drug
  • Use internal standards for each metabolite class to correct for matrix effects

Related Skills

  • metabolomics-analysis/statistical-analysis - General differential analysis
  • metabolomics-analysis/metabolite-annotation - Identify unknown metabolites
  • metabolomics-analysis/xcms-preprocessing - LC-MS preprocessing
  • ms-data-processing/matchms - Spectral matching for metabolite ID

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.