Pathformer: a biological pathway informed transformer for disease diagnosis and prognosis using multi-omics data academic.oup.com/bioinformatics…

SYNY: a pipeline to investigate and visualize collinearity between genomes biorxiv.org/content/10.110…

LucaOne: Generalized Biological Foundation Model with Unified Nucleic Acid and Protein Language biorxiv.org/content/10.110…

New tool announcement! Thanks to a lot of work from Jordan Ort (an awesome student in the lab), with help from Richard Neher, Todd Davis, Tommy Lam, and Sam Shepard, the H5s are now on Nextclade! clades.nextstrain.org A thread below:

Great seminar today by our own Xin Jin, PhD at our Neuroscience Institute Scripps Research. Packed seminar room, exciting unpublished work. So proud of our junior faculty!

Hooray🎉 10 Inspiration4 datasets NASA Open Science Data Repository NASAGeneLab Omics&clinical, first human space data 2B maximally open, findable, accessible, interoperable, reusable 📣 NASASpaceSci Atul Butte Michael Baym Prachee Avasthi Itai Yanai 💔 Bo Wang Katie Link Daniel Kraft, MD

New paper alert! The RNA ribosomal components are so essential that they appear as tons of back to back copies in the genome, which makes them really hard to analyse, so have been mostly ignored. But does it matter how many copies one has? A thread (1/9)

cell.com/cell-genomics/…

New paper introducing RESHAPE, a method simulating genomes of hypothetical descendants that can be used as reference panels safeguarding against re-identification threats. Read more: nature.com/articles/s4358…. Great work done by Théo Cavinato!

What's the economic cost of climate change? Here's the open-access link to our recent nature paper w/Maximilian Kotz Prof Anders Levermann, PhD and - finally - my own short summary below: nature.com/articles/s4158…

So excited to see this released to the world! We've been talking about a new nf-core manuscript for years - in the end it all happend quickly thanks to the collaboration with EuroFAANG!

Have a read and please share widely - this is a great opportunity to reach new audiences!

Introducing: the Deep Tech Database.

A short, but mighty, list of deep tech startups, and the techno-optimistic media pieces that feature them.

🔗— deeptechdb.notion.site —🔗

♒️♒️♒️

Pathway embedding of Pathformer has prevented memory overflow of Transformer module caused by long inputs, but training still requires significant time and space. Therefore, when adding more pathways or gene sets, Pathformer still faces the issue of memory overflow.

Pathformer employs a sparse neural network based on the gene-to-pathway mapping to transform gene embedding into pathway embedding. Pathformer enhances the fusion of information b/n various modalities and pathways by combining pathway crosstalk networks with Transformer encoder.

Pathformer transforms various modalities into distinct gene-level features using a series of statistical methods, such as the maximum value method, and connects these features into a novel compacted multi-modal vector for each gene.

RNAErnie first predicts the possible coarse-grained RNA types using output embeddings and then leverages the predicted types as auxiliary information for fine-tuning. RNAErnie leverages an RNAErnie basic block to predict the top-K most possible coarse-grained RNA types.

RNAErnie model consists of 12 transformer layers. In the motif-aware pretraining phase, RNAErnie is trained on a dataset of approximately 23 million sequences extracted from the RNAcentral database using self-supervised learning with motif-aware multilevel random masking.