Biology subjects

Thompson, I.

Publications and source records attributed to Thompson, I..

4 recordsLinked to original sources

bioRxiv2026

LIMBER: A Hybrid RNA Foundation Model for Predicting and Understanding RNA Structure and Function

RNA foundation models offer a promising route to decoding the relationships between RNA sequence, structure and function, yet how model shapes the acquisition and organization of biological information remains poorly understood. Here we introduce Layer-Interleaved Mamba-Bidirectional Encoder for RNA (LIMBER), a hybrid foundation model integrating bidirectional self-attention with selective state-space modeling, pretrained on 40.5 million non-coding RNA sequences. Through task-specific fine-tuning, LIMBER achieves leading performance on secondary-structure prediction and non-coding RNA classification benchmarks, while demonstrating strong transferability to translation-related tasks and RNA--small-molecule interaction prediction. Beyond model development, we establish a unified, task-agnostic framework to systematically evaluate and interpret the biological knowledge intrinsically acquired by RNA foundation models. This framework validates LIMBER's strong capabilities in recovering RNA structures and predicting mutational effects, while uncovering a hierarchical, task-dependent organization of different biological information across model layers. By further systematically training and evaluating a broad panel of model variants spanning model capacity, training duration and dataset size, we uncover distinct scaling behaviors across structural, evolutionary and functional capabilities. Together, these findings establish LIMBER as a versatile RNA foundation model and provide broader insights into how architectural design, internal representations and scaling shape the learning of RNA sequence-structure-function relationships.

bioinformatics↗

bioRxiv2025

Development and validation of an exome-wide SNP genotyping array for genomic prediction, GWAS and assessment of introgressive hybridization between black and red spruces, and transferability to white and Norway spruces

Introgressive hybridization plays a major role in shaping the evolutionary dynamics and adaptive potential of forest trees. In this study, we developed and validated an exome-wide bispecific SNP genotyping array (Pmr25k) for the closely related species black spruce (Picea mariana) and red spruce (Picea rubens), two ecologically and economically important North American conifers that form a widespread hybrid zone in eastern Canada. Exome capture and sequencing of pooled red spruce samples yielded over 25,000 high-quality SNPs, which were used in conjunction with a previously developed black spruce gene SNP resource of over 97,000 high-quality SNPs, to construct the bispecific genotyping array. The final array comprised 21,573 successfully manufactured SNPs, representing 14,200 distinct gene loci, of which 85% were segregating when both species were considered together. More than 4000 segregating SNPs could also be successfully used and genotyped in each of white spruce (Picea glauca) and Norway spruce (Picea abies), highlighting the conserved nature of DNA attachment sites and presence of homologous SNPs for many gene loci. The Pmr25k array thus provides an efficient and reliable high-throughput genotyping tool to investigate introgression, genetic adaptation at the gene level, and to assist genomic-based prediction for breeding and conservation efforts in boreal spruces.

genomics↗

bioRxiv2025

Age and early life adversity shape heterogeneity of the epigenome across tissues in macaques

Age and early life adversity (ELA) are both key determinants of health, but whether they target similar physiological mechanisms across the body is unknown due to limited multi-tissue datasets from well-characterized cohorts. We generated DNA methylation (DNAm) profiles across 14 tissues in 237 semi-free ranging rhesus macaques, with records of naturally occurring ELA. We show that age-associated DNAm variation is predominantly tissue-dependent, yet tissue-specific epigenetic clocks reveal that the pace of epigenetic aging is relatively consistent within individuals. ELA effects on loci are adversity-dependent, but a given ELA has a coordinated impact across tissues. Finally, ELA targeted many of the same loci as age, but the direction of these effects varied, indicating that ELA does not uniformly contribute to accelerated age in the epigenome. ELA thus imprints a coordinated, tissue-spanning epigenetic signature that is both distinct from and intertwined with age-related change, advancing our understanding of how early environments sculpt the molecular foundations of aging and disease.

genomics↗

bioRxiv2018

Characterisation and development of aspirin inducible biosensors in E. coli Nissle 1917 and SimCells

A simple aspirin-inducible system has been developed by employing the Psal promoter and SalR regulation system originally from Acinetobacter baylyi ADP1, which has been cloned into E. coli for characterisation of gene circuits and induction of novel SimCells (simple cells). Mutagenesis at the DNA binding domain (DBD) and chemical recognition domain (CRD) of the SalR protein in A. baylyi ADP1 suggests that inactive SalRi can compete with activated SalRa, occupying the binding position of Psal promoter. The induction of the Psal promoter was compared in two different designs in E. coli: simple regulation (SRS) and positive autoregulated system (PAR). Both regulatory systems were induced in a dose-dependent manner in the presence of aspirin in the range of 0.05-10 M. Over-expression of SalR in the SRS system reduces both baseline leakiness and inducible strength of Psal promoter. A weak SalR expression significantly improve the inducible strength, which is in a good agreement of the proposed hypothesis of SalRi/SalRa competitive binding. The PAR system provides a feedback loop that fine-tunes the level of SalR, displaying inducible strength. A mathematical model based on SalRi/SalRa competitive binding hypothesis was developed, which not only reproduces the observed experimental results but also predict the performance of a new gene circuit design. The aspirin-inducible systems were also functional in probiotic strain E.coli Nissle 1917 (EcN) and SimCells produced from E. coli MC1000 {Delta}minD. The well-characterised and modularised aspirin-inducible gene circuits would be useful biobricks for bacterial therapy in environment and medical applications.

microbiology↗