Biology subjects

Schmidlin, T.

Publications and source records attributed to Schmidlin, T..

3 recordsLinked to original sources

bioRxiv2026

The splicing paralogues SNRPB and SNRPN control differential metabolic states.

Gene duplication underlies evolutionary innovation, yet many paralogues remain highly similar, raising questions about their functional divergence and physiological relevance. The spliceosomal Sm core protein SNRPB and its mammalian-specific paralogue SNRPN share over 90% sequence identity, but their distinct expression patterns - SNRPB being ubiquitous and SNRPN confined to the brain - suggest specialized functions. Why mammals have two different spliceosomes has remained obscure. Here, we generated isogenic human cell lines expressing ectopically either SNRPB or SNRPN exclusively and found that SNRPN stabilizes transcripts involved in energy metabolism and mitochondrial function, leading to increased mitochondrial abundance and oxygen consumption. Despite similar spliceosomal interactomes, SNRPN more strongly associates with the PRMT5 methylosome complex and exhibits dynamic arginine methylation in its C-terminal region that is sensitive to translation inhibition and amino acid availability. The SNRPN-dependent transcriptome responds to translation inhibition by stabilizing long, intron-rich genes involved in amino acid and energy metabolism. Our findings reveal a nutrient-sensitive, methylation-dependent mechanism that differentiates the two paralogues. This suggests that SNRPN functions as a metabolic-specialized spliceosomal subunit thereby providing tissue-specific adaptation of RNA processing in mammals.

molecular biology↗

bioRxiv2026

Cross-Domain Transfer Learning from Peptides to Lipids Using a Multi-Property Fine-Tuned LLM

Accurate knowledge of liquid chromatography retention time (RT) is essential for confident compound identification in metabolomics and lipidomics. Yet, it is often constrained by the scarcity of experimental data for many molecular classes. Current workflows depend on experimental RT libraries, which are time-consuming to build and limited to previously observed compounds. Here, we present a transfer learning pipeline that leverages large, publicly available peptide datasets to enable accurate lipid RT prediction in data-sparse scenarios. We first show that a ChemBERTa language model, when pre-trained on peptides with a multi-task objective (predicting both RT and fundamental RDKit molecular descriptors), learns a more robust and generalizable chemical representation than a single-task (RT-only) model. This multi-property pre-training yielded superior generalization in lipids, achieving test R2 values of 0.842 against 0.814 (RT-only). Crucially, transferring this peptide-based model to lipid data provided a pronounced advantage in data-sparse scenarios. When fine-tuned on only 5% of available lipid data, the transferred model improved the median test R2 by +0.234 over a model trained from scratch. Significant benefits persisted at intermediate data scales (50-75%), with performance converging only when 100% of the lipid data was used. Notably, the pre-trained model never underperformed the baseline, exhibiting more stable training across all data scales. These results demonstrate that multi-property pre-training guides language models towards chemically meaningful representations that support better RT prediction in different molecular domains. Furthermore, peptide-based pre-training facilitates cross-domain transfer of chemical properties to lipid. Our work provides a practical, scalable strategy to mitigate data scarcity in lipidomics by transferring knowledge from data-rich peptide databases, offering a computational alternative to extensive experimental library generation and enabling more confident identification in small-scale omics studies.

bioinformatics↗

bioRxiv2020

Genomic stability and adaptation of beer brewing yeasts during serial repitching in the brewery

Ale brewing yeast are the result of admixture between diverse strains of Saccharomyces cerevisiae, resulting in a heterozygous tetraploid that has since undergone numerous genomic rearrangements. As a result, comparisons between the genomes of modern related ale brewing strains show both extensive aneuploidy and mitotic recombination that has resulted in a loss of intragenomic diversity. Similar patterns of intraspecific admixture and subsequent selection for one haplotype have been seen in many domesticated crops, potentially reflecting a general pattern of domestication syndrome between these systems. We set out to explore the evolution of the ale brewing yeast, to understand both polyploid evolution and the process of domestication in the ecologically relevant environment of the brewery. Utilizing a common brewery practice known as repitching, in which yeasts are reused over multiple beer fermentations, we generated population time courses from multiple breweries utilizing similar strains of ale yeast. Applying whole-genome sequencing to the time courses, we have found that the same structural variations in the form of aneuploidy and mitotic recombination of particular chromosomes reproducibly rise to detectable frequency during adaptation to brewing conditions across multiple related strains in different breweries. Our results demonstrate that domestication of ale strains is an ongoing process and will likely continue to occur as modern brewing practices develop.

evolutionary biology↗