Machine Learning-Corrected Isotope Ratio Data for Metabolic Flux Analysis

A dataset supporting a machine learning correction algorithm for isotope ratio measurements in metabolic labeling studies using Orbitrap mass spectrometry. The dataset, authored by Zhenwen Yu and last updated in May 2026, is a 141.3 MB ZIP file. It likely contains scan-by-scan mass spectrometry data used to train a random forest model for bias correction.

Use Cases

• Training bias-correction models for mass isotopomer ratios based on scan-level features like ion signal intensity.
• Ranking factors impacting measurement bias in untargeted mass spectrometry based on the model's feature importance output.
• Improving the accuracy of metabolic flux measurements from heavy water labeling studies using the provided correction method.
• Enabling high-throughput, untargeted 'fluxomics' experiments by overcoming Orbitrap accuracy limitations across a wide m/z range.

Strengths

• The associated random forest model reportedly reduces mean absolute percentage error for the M1 mass isotopomer from 21.3% to 3.5%.
• The model identifies and ranks bias-causing factors, highlighting the ion signal intensity/TIC ratio as a previously neglected dominant factor.
• Dataset is shared under a CC-BY-NC-4.0 license, permitting non-commercial reuse with attribution.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for specific modeling tasks.
• The description focuses on the method and results; the exact structure and content of the data files require inspection.

Provenance

Source

figshare

Collection Method

Likely generated from Orbitrap mass spectrometry experiments in metabolic labeling studies.

Time Range

null

Freshness

Last updated 2026-05-04 14:57:13; freshness should be verified.

Geography

null