molecular-property-profiling - SKILL.md Agent Skill

name: molecular-property-profiling description: Comprehensive molecular property analysis covering basic info, hydrophobicity, H-bonding, structural complexity, topology, drug-likeness, charge distribution, and complexity metrics. license: MIT license metadata: skill-author: PJLab

Molecular Property Profiling Workflow

Usage

1. MCP Server Definition

Use the same DrugSDAClient class as defined in previous skills.

2. Comprehensive Molecular Property Analysis

This workflow computes a comprehensive set of molecular descriptors across 8 different categories, providing a complete molecular profile for QSAR modeling, drug discovery, and molecular analysis.

Workflow Steps:

Basic Properties - Molecular formula, weight, atom counts, bond counts
Hydrophobicity - LogP, molar refractivity, lipophilicity descriptors
Hydrogen Bonding - H-bond donors/acceptors, TPSA
Structural Complexity - Ring counts, aromatic rings, rotatable bonds
Topological Descriptors - Chi indices, Kappa shape indices
Drug Chemistry - QED score, Lipinski violations
Charge Properties - Gasteiger charges, formal charge
Complexity Metrics - Molecular complexity, asphericity

Implementation:

from collections import defaultdict

def merge_lists_by_smiles(*lists):
    """Merge multiple descriptor lists by SMILES key"""
    merged = defaultdict(dict)
    for lst in lists:
        for d in lst:
            smiles = d['smiles']
            merged[smiles].update(d)
    return list(merged.values())

client = DrugSDAClient("https://scp.intern-ai.org.cn/api/v1/mcp/2/DrugSDA-Tool")
if not await client.connect():
    print("connection failed")
    return

## Input: List of SMILES strings
smiles_list = [
    'Nc1nnc(S(=O)(=O)NCCc2ccc(O)cc2)s1',
    'COc1ccc2c(=O)cc(C(=O)N3CCN(c4ccc(F)cc4)CC3)oc2c1',
    'CCCC1CCC(CC(=O)Cl)(C2CCCCC2)CC1'
]

## Step 1: Calculate basic molecular properties
result = await client.session.call_tool(
    "calculate_mol_basic_info",
    arguments={"smiles_list": smiles_list}
)
basic_metrics = client.parse_result(result)['metrics']

## Step 2: Calculate hydrophobicity descriptors
result = await client.session.call_tool(
    "calculate_mol_hydrophobicity",
    arguments={"smiles_list": smiles_list}
)
hydrophobicity_metrics = client.parse_result(result)['metrics']

## Step 3: Calculate hydrogen bonding properties
result = await client.session.call_tool(
    "calculate_mol_hbond",
    arguments={"smiles_list": smiles_list}
)
hbond_metrics = client.parse_result(result)['metrics']

## Step 4: Calculate structural complexity
result = await client.session.call_tool(
    "calculate_mol_structure_complexity",
    arguments={"smiles_list": smiles_list}
)
structure_metrics = client.parse_result(result)['metrics']

## Step 5: Calculate topological descriptors
result = await client.session.call_tool(
    "calculate_mol_topology",
    arguments={"smiles_list": smiles_list}
)
topology_metrics = client.parse_result(result)['metrics']

## Step 6: Calculate drug chemistry properties
result = await client.session.call_tool(
    "calculate_mol_drug_chemistry",
    arguments={"smiles_list": smiles_list}
)
chemistry_metrics = client.parse_result(result)['metrics']

## Step 7: Calculate charge properties
result = await client.session.call_tool(
    "calculate_mol_charge",
    arguments={"smiles_list": smiles_list}
)
charge_metrics = client.parse_result(result)['metrics']

## Step 8: Calculate complexity metrics
result = await client.session.call_tool(
    "calculate_mol_complexity",
    arguments={"smiles_list": smiles_list}
)
complexity_metrics = client.parse_result(result)['metrics']

## Merge all descriptors by SMILES
complete_profiles = merge_lists_by_smiles(
    basic_metrics,
    hydrophobicity_metrics,
    hbond_metrics,
    structure_metrics,
    topology_metrics,
    chemistry_metrics,
    charge_metrics,
    complexity_metrics
)

## Display results
for profile in complete_profiles:
    print(f"\nSMILES: {profile['smiles']}")
    print(f"Molecular Formula: {profile['molecular_formula']}")
    print(f"Molecular Weight: {profile['molecular_weight']:.2f}")
    print(f"LogP: {profile['logp']:.2f}")
    print(f"QED Score: {profile['qed']:.4f}")
    print(f"H-Bond Donors: {profile['num_h_donors']}")
    print(f"H-Bond Acceptors: {profile['num_h_acceptors']}")
    print(f"TPSA: {profile['tpsa']:.2f}")
    print(f"Lipinski Violations: {profile['lipinski_rule_of_5_violations']}")

await client.disconnect()

Descriptor Categories

1. Basic Properties

molecular_formula: Molecular formula
molecular_weight: Molecular weight (Da)
num_heavy_atoms: Count of non-hydrogen atoms
num_atoms, num_bonds: Total atom and bond counts
formal_charge: Overall formal charge

2. Hydrophobicity

logp: Partition coefficient (lipophilicity)
molar_refractivity: Molar refractivity
fraction_csp3: Fraction of sp3 carbons (saturation)

3. Hydrogen Bonding

num_h_donors: H-bond donor count
num_h_acceptors: H-bond acceptor count
tpsa: Topological polar surface area (Ų)

4. Structural Complexity

num_rings, num_aromatic_rings: Ring counts
num_rotatable_bonds: Flexible bonds
num_heteroatoms: Non-C/H atoms

5. Topological Descriptors

chi0v-chi4v: Chi connectivity indices
kappa1-kappa3: Kappa shape indices
hall_kier_alpha: Hall-Kier alpha value

6. Drug Chemistry

qed: Quantitative Estimate of Drug-likeness (0-1)
lipinski_rule_of_5_violations: Lipinski violations (0-4)

7. Charge Properties

min/max/avg_gasteiger_charge: Gasteiger partial charges
gasteiger_charge_range: Charge distribution range

8. Complexity Metrics

molecular_complexity: Bertz complexity index
aromatic_proportion: Fraction of aromatic atoms
asphericity: 3D shape asphericity

Input/Output

Input:

smiles_list: List of SMILES strings

Output:

List of dictionaries, each containing 50+ molecular descriptors for one molecule

Applications

QSAR Modeling: Use descriptors as features for predictive models
Drug Discovery: Screen compounds by drug-likeness and physicochemical properties
Chemical Space Analysis: Visualize and cluster molecules by properties
Lead Optimization: Track property changes during optimization
Virtual Screening: Filter libraries by desired property ranges

Property Filters for Drug-likeness

Typical ranges for oral drug candidates:

Molecular Weight: 150-500 Da
LogP: 0-5
H-Bond Donors: ≤ 5
H-Bond Acceptors: ≤ 10
TPSA: 20-140 Ų
Rotatable Bonds: ≤ 10
QED Score: > 0.5