The way researchers standardize multiple reaction monitoring (MRM) lipid data can change which biological differences appear in the results, a methods paper shows. In a neonatal piglet-liver validation dataset, four approaches produced markedly different sets of lipids flagged as different between groups.
The effect was visible in a broad sample-level analysis. Principal component analysis showed more apparent separation and tighter clustering between the two groups when proportion-based approaches were used than when the data were handled in concentration-like form.
One dataset, four ways to measure it
The tutorial is built around a semitargeted MRM workflow. It starts with a discovery list corresponding to approximately 5,000 lipid species, then screens in individual samples only the transitions detected in pooled samples.
It compares four standardization options: raw intensity, scaling to an internal standard, relative abundance within a lipid class, and sample injection per MRM list. The first two are concentration-like approaches; the latter two describe lipid proportions.
The reported workflow used blank-based filters at different stages. Transitions with raw ion intensity 1.3-fold above the water blank were selected for individual screening, producing 906 MRM transitions. A 1.5-fold-above-blank filter retained 615 MRM signals. Triglyceride signals accounted for 40% of the retained lipid set, followed by PC at 16%, PE at 9% and DG at 8%.
The paper also notes that different fatty-acyl-chain neutral-loss signals can make the same triglyceride appear more than once in the retained data, an important feature when interpreting the lipid counts.
The statistical answer depended on the method
At the study’s false-discovery-rate cutoff of less than 0.05, the four approaches flagged 132 raw-intensity, 83 internal-standard, 261 MRM-list and 217 class-standardized differential lipids. Only 50 of the 691 evaluated lipids were common to all four lists.
The differences extended beyond the size of the lists. Among the 50 common lipids, concentration-like methods produced larger log2 fold-change magnitudes than proportion-like methods.
The contrast was especially strong between the two proportion-based approaches. To include all findings from the MRM-list method, the class-standardization false-discovery-rate cutoff had to be relaxed from 0.05 to 0.98. Multiple listed classes had class-based false-discovery rates above 0.25, and the text describes the MRM-list approach as highly sensitive.
What emerged in piglet liver
The validation compared eight ZH liver samples, from piglets euthanized immediately after birth, with nine SOS samples, from piglets suckled by the dam for 24 hours and then fasted for 2 hours.
A Lipid Ontology analysis of class-sum-standardized data found SOS-associated enrichment of descriptors related to mitochondria, lipid storage, lipid droplets and negatively charged headgroups. The analysis ranked lipid identifiers with a one-tailed t-test and tested enrichment with a one-tailed low-to-high Kolmogorov-Smirnov test.
The triglyceride patterns also differed. SOS had more longer-chain triglycerides with 52 to 57 total carbons and more polyunsaturated triglycerides and fatty acids, including C18:2, a fatty acid notation indicating 18 carbons and two double bonds. ZH had more triglycerides with 44 to 51 total carbons and more saturated or monounsaturated fatty acids.
A useful guide with a narrow reach
This was a descriptive comparison, not a causal test. The two groups differed in postnatal timing, feeding, fasting and euthanasia timing, so the design cannot isolate one factor or establish that suckling caused the observed liver-lipid differences.
The measurement design also sets limits. MRM is semitargeted and works from predefined transitions, while sum-based standardization produces compositional proportions rather than absolute lipid amounts. The results therefore do not establish that one standardization approach is universally superior.
The authors’ practical message is to match the standardization method to the biological question. The tutorial offers a workflow for researchers analyzing semitargeted MRM lipid data, but its biological findings remain specific to this piglet-liver panel, the reported thresholds and the two-group validation comparison.
Paper data and sources
Original title: Step-by-step tutorial for analysis of Multiple Reaction Monitoring Profiling of lipidome data for biological interpretation
Authors: Not provided
Journal/Repository: Journal of biomolecular techniques : JBT
Status: Peer-reviewed
First online: 2026-08-19
DOI: Not available
Original paper · Full text