agentleFS
dailycafidailycafi/metabolism-skills/skills/systems-biology/context-specific-models/SKILL.md

bio-systems-biology-context-specific-models

Skill7 starsChanged 7 months agoMITView on GitHub

Build tissue-specific and condition-specific metabolic models by integrating gene expression data with GEMs using GIMME, iMAT, and INIT. Use when: user wants tissue-specific metabolism modeling, needs to constrain a metabolic model with RNA-seq data, or create context-specific FBA models. Triggers: tissue-specific metabolism, GIMME, iMAT, INIT, context-specific model, expression-constrained FBA, cell-type metabolism, condition-specific metabolic model, transcriptomics + FBA, liver/brain/cancer metabolism model.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    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 16
  2. Distinctive:

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

    line 19

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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: COBRApy 0.29+, numpy 1.26+, pandas 2.2+

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

  • Python: pip show <package> then help(module.function) to check signatures

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

Context-Specific Models

"Build a tissue-specific metabolic model from my expression data" → Constrain a generic genome-scale model using transcriptomics data to produce a context-specific model reflecting the active metabolism of a particular tissue or condition, using GIMME, iMAT, or INIT algorithms.

  • Python: custom implementations with cobra model manipulation (COBRApy)

GIMME Algorithm

Goal: Build a tissue-specific metabolic model by integrating transcriptomics data with a generic genome-scale model, retaining only metabolically active reactions.

Approach: Map gene expression values to reactions, penalize flux through lowly-expressed reactions while maintaining minimum biomass production, and remove inactive reactions to produce a context-specific model.

import cobra
import numpy as np

def gimme(model, expression_data, threshold=0.25, required_growth=0.1):
    '''Gene Inactivity Moderated by Metabolism and Expression (GIMME)

    Creates context-specific model by:
    1. Penalizing flux through lowly-expressed reactions
    2. Requiring minimum biomass production

    Args:
        expression_data: dict mapping gene_id -> expression value
        threshold: Expression percentile below which genes are inactive
                  0.25 = bottom 25% considered inactive
        required_growth: Minimum growth rate to maintain

    Returns:
        Context-specific model with inactive reactions constrained
    '''
    # Calculate expression threshold
    values = list(expression_data.values())
    cutoff = np.percentile(values, threshold * 100)

    # Identify lowly-expressed genes
    low_expressed = {g for g, v in expression_data.items() if v < cutoff}

    # Create context model
    context_model = model.copy()

    # Set minimum growth constraint
    context_model.reactions.get_by_id('Biomass_Ecoli_core').lower_bound = required_growth

    # Minimize flux through reactions with low-expressed genes
    for rxn in context_model.reactions:
        genes = {g.id for g in rxn.genes}
        if genes and genes.issubset(low_expressed):
            # This reaction is likely inactive - constrain it
            rxn.upper_bound = min(rxn.upper_bound, 1.0)
            rxn.lower_bound = max(rxn.lower_bound, -1.0)

    return context_model

iMAT Algorithm

def imat(model, expression_data, high_threshold=0.75, low_threshold=0.25):
    '''Integrative Metabolic Analysis Tool (iMAT)

    Maximizes agreement between flux activity and expression:
    - Highly expressed reactions should carry flux
    - Lowly expressed reactions should have zero flux

    More sophisticated than GIMME - uses MILP optimization.
    '''
    from cobra import Reaction

    # Classify reactions by expression
    high_expr_rxns = []
    low_expr_rxns = []

    for rxn in model.reactions:
        if rxn.genes:
            # Aggregate gene expression (use max for OR, min for AND)
            gene_expr = [expression_data.get(g.id, 0.5) for g in rxn.genes]
            rxn_expr = max(gene_expr)  # Simplified OR logic

            if rxn_expr > np.percentile(list(expression_data.values()), high_threshold * 100):
                high_expr_rxns.append(rxn.id)
            elif rxn_expr < np.percentile(list(expression_data.values()), low_threshold * 100):
                low_expr_rxns.append(rxn.id)

    # Create MILP to maximize consistent reactions
    # This is a simplified version - full iMAT uses binary variables
    context_model = model.copy()

    # Force flux through highly expressed reactions
    for rxn_id in high_expr_rxns:
        rxn = context_model.reactions.get_by_id(rxn_id)
        rxn.lower_bound = max(rxn.lower_bound, 0.01)

    # Constrain lowly expressed reactions
    for rxn_id in low_expr_rxns:
        rxn = context_model.reactions.get_by_id(rxn_id)
        rxn.upper_bound = min(rxn.upper_bound, 0.1)
        rxn.lower_bound = max(rxn.lower_bound, -0.1)

    return context_model, high_expr_rxns, low_expr_rxns

Expression Data Integration

def load_expression_data(filepath, gene_col='gene_id', expr_col='TPM'):
    '''Load and normalize expression data

    Accepts:
    - RNA-seq counts (TPM, FPKM)
    - Microarray intensities
    - Proteomics abundances

    Returns dict mapping gene_id -> normalized expression
    '''
    import pandas as pd

    df = pd.read_csv(filepath)

    # Log-transform if needed (high dynamic range)
    expr = df[expr_col].values
    if expr.max() / expr.mean() > 100:
        expr = np.log2(expr + 1)

    # Normalize to 0-1 range
    expr_norm = (expr - expr.min()) / (expr.max() - expr.min())

    return dict(zip(df[gene_col], expr_norm))


def aggregate_gene_expression(model, expression_data, method='max'):
    '''Map gene expression to reactions

    Methods:
    - 'max': Use maximum gene expression (OR logic)
    - 'min': Use minimum gene expression (AND logic)
    - 'mean': Average across genes

    For GPR: (A and B) or C
    - min(A, B) for the complex
    - max(complex, C) for the alternatives
    '''
    rxn_expression = {}

    for rxn in model.reactions:
        if not rxn.genes:
            rxn_expression[rxn.id] = 0.5  # Default for non-enzymatic
            continue

        gene_expr = [expression_data.get(g.id, 0.5) for g in rxn.genes]

        if method == 'max':
            rxn_expression[rxn.id] = max(gene_expr)
        elif method == 'min':
            rxn_expression[rxn.id] = min(gene_expr)
        else:
            rxn_expression[rxn.id] = np.mean(gene_expr)

    return rxn_expression

Tissue-Specific Human Models

def create_tissue_model(generic_model, gtex_expression, tissue='liver'):
    '''Create tissue-specific model from GTEx expression data

    GTEx provides median TPM for 54 human tissues.
    Download from: https://gtexportal.org/home/datasets
    '''
    import pandas as pd

    # Load GTEx median expression
    gtex = pd.read_csv(gtex_expression, sep='\t')

    # Extract tissue column
    tissue_col = [c for c in gtex.columns if tissue.lower() in c.lower()][0]
    expression = dict(zip(gtex['gene_id'], gtex[tissue_col]))

    # Apply GIMME
    tissue_model = gimme(generic_model, expression, threshold=0.25)

    return tissue_model

Validate Context Model

def validate_context_model(original, context, expression_data):
    '''Compare original and context-specific models

    Checks:
    1. Growth capability maintained
    2. Inactive reactions reduced
    3. Active reactions maintained
    '''
    # Growth comparison
    orig_growth = original.optimize().objective_value
    context_growth = context.optimize().objective_value

    # Count constrained reactions
    constrained = 0
    for rxn in context.reactions:
        orig_rxn = original.reactions.get_by_id(rxn.id)
        if rxn.upper_bound < orig_rxn.upper_bound:
            constrained += 1

    return {
        'original_growth': orig_growth,
        'context_growth': context_growth,
        'growth_ratio': context_growth / orig_growth,
        'constrained_reactions': constrained,
        'total_reactions': len(context.reactions)
    }

Related Skills

  • systems-biology/flux-balance-analysis - Run FBA on context models
  • differential-expression/de-results - Generate expression data
  • single-cell/clustering - Cell-type specific expression

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.