Dehui Lin

HERRO - imagine what you can do with accurate ultra-long nanopore reads biorxiv.org/content/10.110… w/ @domstanojevic @DehuiLin @PaolaFlorezdeS and great help from @sergeynurk #nanopore. See you at London Calling 2024!!

Our new paper is out! We made a lot of progress on our GNN-based de novo genome assembly paradigm and here we present all our findings and progress 🧬🥳 Code: github.com/lbcb-sci/GNNome Paper: biorxiv.org/content/10.110… See details of GNNome below 👇🧵

Happy to present RiNALMO - our RNA large language model arxiv.org/abs/2403.00043 w/ @RJPenic Tin Vlasic @ywan_wan and Roland Huber. RiNALMo is the largest RNA language model to date, with 650 million parameters pre-trained on 36 million non-coding RNA sequences. 1/2

Are nanopore UL reads only long reads we need? We developed Herro github.com/lbcb-sci/herro AI error correction model that can correct reads to accuracy above Q30 while trying to keep informative positions intact. w/ @domstanojevic @DehuiLin @sergeynurk @PaolaFlorezdeS @nanopore

Long reads metagenome benchmark is out. Highlights. 1. In most cases Kraken, 2. minimap2/ram for slightly higher accuracy. 3. The right database is of huge importance 4. Check taxonomy files carefully. bmcbioinformatics.biomedcentral.com/articles/10.11… w/@NiranjanTW @KrizanovicK @jmaricb @Sylvain14518009

After many months, I'm proud to share our work on untangling genome assembly graphs with GNNs. Or, GNNome Assembly (sorry not sorry)🧬 Paper: arxiv.org/abs/2206.00668 Code/Data: github.com/lvrcek/GNNome-… with @xbresson, T. Laurent, @Martin_fschmitz, and @msikic. Read more in 🧵