Including data from 1,047 patients across 19 inflammatory diseases, a new atlas presents a comprehensive model of inflammation in circulating immune cells.
nature.com/articles/s41591-0…
Protein large language model assisted one-to-one gene homology mapping in cross-species single-cell transcriptome integration
1. A new study presents a novel approach to cross-species single-cell transcriptome integration by improving gene homology mapping, which is crucial for comparative analysis of gene expression profiles across different species.
2. The traditional Ensembl homology tables often lead to many-to-many mappings that can introduce noise and complicate biological interpretation. This study proposes a one-to-one mapping strategy using a protein large language model (pLLM) to enhance accuracy and biological relevance.
3. The pLLM-based method integrates sequence similarity with high-dimensional embeddings derived from the model, resulting in a fused mapping approach that significantly outperforms existing methods in a comprehensive benchmark across nine datasets, 11 species, and over 3.2 million cells.
4. The study demonstrates that the one-to-one mapping strategies effectively reduce gene-family-driven noisy clusters and improve species mixing and cell-type separation, leading to clearer biological signals.
5. Additionally, the pLLM-based approach uncovers previously unannotated cross-species marker-gene pairs, including those between evolutionarily distant species, facilitating novel discoveries in comparative genomics.
6. The fused mapping strategy, HL_O2O, combines the strengths of both sequence-based homology and pLLM-derived embeddings, recovering biologically meaningful gene pairs that are typically lost under strict homology filters.
7. The findings highlight the importance of integrating sequence and representation-based evidence for accurate cross-species single-cell transcriptome integration, providing a robust framework for future comparative biological studies.
📜Paper: biorxiv.org/content/10.1101/…#ComputationalBiology#GeneHomology#SingleCellTranscriptomics#CrossSpeciesIntegration#ProteinLanguageModel
I am thrilled to share our work published today
@NatImmunol. We describe transcriptional changes of endothelial and immune cell subsets from brain and blood after stroke. Read it here: rdcu.be/du8UE. Check our tool for easy data exploration: anratherlab.shinyapps.io/str…
🧬 Wrapping up an incredible 4-day journey at #ILS Bhubaneshwar! 🚀 Our hands-on workshop on single-cell transcriptomics with BCR/TCR profiling using the 10x Genomics platform was a resounding success. 🎉👋 #SingleCellTranscriptomics#GenomicsWorkshop#ResearchSuccess
We are looking for two computational scientists to join @XiaoFu90's Integrative Modelling lab and to support their investigations into organisational principles and dynamics of the tumour microenvironment in solid tumours.
Find out more: beatson.gla.ac.uk/careers/po…