Skip to content

BioPAN

Summary

BioPAN interprets quantitative lipidomic differences between two conditions as changes in curated mammalian biosynthetic reactions. It produces interactive lipid-reaction graphs, scores activated or suppressed reactions and pathways, and proposes genes encoding relevant enzymes, turning a lipid abundance table into hypotheses for follow-up with transcriptomic or proteomic data [1–3].

Inputs and Processing

BioPAN accepts a CSV file with lipid identifiers in LIPID MAPS shorthand and quantitative columns for at least two replicates per condition; three or more replicates are recommended. LipidLynxX is used to harmonize some submitted names, after which lipids are classified as recognized/processed, recognized but unprocessed, or unrecognized [1].

Only lipids that can be assigned to at least one curated reaction proceed to pathway analysis. In the 2021 publication the manually assembled mammalian database contained 94 reactions spanning 41 lipid subclasses, so an unprocessed result can reflect pathway coverage or missing matched substrate/product composition rather than poor data quality [1].

Scoring and Output

Reaction Z-scores compare product and substrate changes between conditions; a cumulative function combines reaction evidence into pathway scores. By default, BioPAN labels reactions or pathways as modified at p < 0.05, corresponding to Z > 1.645 in the method [1].

The results interface provides:

  • interactive subclass- or molecular-species-level networks;
  • active and suppressed reaction/pathway views;
  • filtering by lipid subset and significance threshold;
  • tables of reaction chains, scores, and candidate genes;
  • links from lipid nodes to LMSD and from genes to LMPD;
  • export of graphs and tabular results [1].

Interpretation Boundaries

BioPAN scores consistency with a curated reaction network; it does not directly measure enzyme activity or establish that a proposed gene caused the lipid change. Gene lists are mechanistic candidates, and species must be checked because the default gene nomenclature is human even when the input data come from another mammal [1].

The published example used young-versus-aged mouse liver and cerebral cortex data. BioPAN reproduced reported ether-linked diacylglycerol accumulation and suggested tissue-specific differences in triacylglycerol and sphingolipid pathways, illustrating hypothesis generation rather than independent validation [1].

Availability

The web application, documentation, and tutorial are hosted by LIPID MAPS. The publication archived source materials under GNU GPL 3.0 [1].

Citations

[1] Gaud et al. 2021. BioPAN: a web-based tool to explore mammalian lipidome metabolic pathways on LIPID MAPS, version 2. Supports: Input format, harmonization, curated-reaction coverage, scoring, interface, example analysis, limitations, and availability. Location: Methods—Implementation and Operation; Example Use Case; Conclusion; Software Availability.

[2] Conroy et al. 2024. LIPID MAPS: update to databases and tools for the lipidomics community. Supports: Integration with LMSD reaction data and current LIPID MAPS deployment. Location: Updates to LMSD; Tools.

[3] Fedorova et al. 2023. Guiding the choice of informatics software and tools for lipidomics research applications. Supports: BioPAN's role among pathway and network integration tools. Location: Data Integration Solutions.