Bio Metabolomics Isotope Tracing

Version Compatibility

Reference examples tested with: isocor 2.2+, numpy 1.26+

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.

Isotope Tracing / Stable-Isotope-Resolved Metabolomics

"What is this pathway actually doing, not just how much metabolite is there?" -> Feed a labeled tracer, then measure how label propagates into downstream metabolites as a mass-isotopomer distribution (MID) over time.

• Python: isocor.mscorrectors.MetaboliteCorrectorFactory().correct() (IsoCor) for natural-abundance correction
• R: accucor::natural_abundance_correction() (AccuCor) for high-resolution correction
• Modeling layer (separate discipline): INCA, 13CFLUX2, OpenFLUX for 13C-MFA / INST-MFA flux fitting

The Single Most Important Insight -- Labeling Reports Flux; Pool Size Does Not

A metabolite's concentration is how much is there; its labeling pattern (MID) is where the carbon came from and how fast it got there. These are independent measurements and frequently move in OPPOSITE directions: block a downstream-consuming enzyme and the intermediate pool rises (it backs up) while the labeling of downstream products falls (flux through them dropped). Reading the pool alone reports the opposite of the biology. An isotope-tracing experiment therefore answers a fundamentally different question than untargeted or targeted abundance profiling, and its analysis is dominated by two mandatory corrections that fabricate flux if skipped: natural-abundance / tracer-purity correction (raw isotopologue areas are NOT the labeling), and the steady-state assumption (a single MID is a snapshot whose meaning depends on whether labeling has plateaued). Fractional enrichment is concentration-independent (it is a ratio within one pool), which is why it survives the recovery/matrix problems that plague absolute quant -- but it says nothing about amount.

Core Concepts

Decision Tree by Scenario

Natural-Abundance + Tracer-Purity Correction (mandatory)

Goal: Turn raw measured isotopologue areas into a true MID that reflects only tracer-derived label.

Approach: Even a fully unlabeled molecule shows an M+1, M+2 ladder because ~1.07% of carbon is naturally 13C (plus 15N, 2H, 18O, 34S, and derivatization Si). Build the natural-abundance ladder from the molecular (and derivative) formula, deconvolve it out, then correct for the tracer not being 100% isotopically pure. Feeding uncorrected areas to a flux model is the equivalent of reporting an uncalibrated peak area as a concentration.

import isocor

# corrector knows the formula's natural-abundance ladder and the tracer
corrector = isocor.mscorrectors.MetaboliteCorrectorFactory(
    'C6H12O6', tracer='13C',
    correct_NA_tracer=True,           # also strip the labeled element's own natural abundance
    tracer_purity=[0.01, 0.99])       # [unlabeled, labeled] per-position purity of the tracer

# raw measured areas M+0..M+6 for a partially labeled glucose pool
corrected_area, iso_fraction, residuum, mean_enrichment = corrector.correct(
    [50000., 8000., 12000., 3000., 1500., 6000., 25000.])
# iso_fraction is the corrected MID; mean_enrichment is fractional enrichment

High-resolution Orbitrap data resolves 13C from 15N/2H by exact mass, enabling a different (often simpler) correction; AccuCor (R) is tuned for that case:

library(accucor)
# El-MAVEN / MAVEN isotopologue table; Resolution is the instrument resolving power
corrected <- natural_abundance_correction(path = 'elmaven_export.xlsx',
                                          resolution = 100000, purity = 0.99)

Pick the corrector by tracer count and resolution: IsoCor handles any tracer at any resolution; AccuCor (single tracer) and AccuCor2 (dual 13C-15N / 13C-2H) target high-res. Verify the chosen tool's current argument names before running -- both APIs drift across versions.

Computing and Plotting an MID / Fractional Enrichment

Goal: Summarize a corrected isotopologue vector as an MID and one fractional-enrichment number, comparably across conditions.

Approach: Normalize corrected areas to sum 1 (the MID), then take the atom-weighted mean over isotopologue index divided by the number of tracer atoms.

import numpy as np

corrected = np.array([26000., 2200., 5600., 1200., 500., 2300., 12500.])
mid = corrected / corrected.sum()                              # M+0..M+n fractions
fractional_enrichment = np.sum(np.arange(len(mid)) * mid) / (len(mid) - 1)
# stacked-bar MID per condition is the standard visualization; never plot raw (uncorrected) areas

Steady-State Check (does the labeling number mean anything yet?)

Goal: Decide whether a measured MID may be read as flux-informative or is still a kinetic transient.

Approach: Sample labeling at several timepoints; isotopic steady state is reached when the MID stops changing (plateau). Only plateau MIDs license classical-MFA flux inference; a rising MID is kinetic data requiring INST-MFA.

import numpy as np

# fractional enrichment per timepoint (minutes) for one metabolite
t = np.array([0, 5, 15, 30, 60, 120])
fe = np.array([0.00, 0.18, 0.31, 0.39, 0.42, 0.43])
reached_plateau = abs(fe[-1] - fe[-2]) < 0.02     # <2% change between last points = plateau
# if not reached_plateau: the pool is still labeling -> use the full time course (INST-MFA), not one point

Per-Method Failure Modes