Konrad Krawczyk

A new pre-print from our team demonstrated that pre-training on large set of antibody models allows producing #anitbody physically plausible shapes on an artificial set of #CDR-H3s, showing the ability to generalize to the vast #antibodysequence space naturalantibody.com/use-case/struc…

Our preprint on OpenFold, our trainable reproduction of AlphaFold2, is finally up (biorxiv.org/cgi/content/sh…)! Since we open-sourced parameters in June, we've trained the model to high accuracy more than 50 times, on a variety of datasets. Here's what we learned (a lot) -> (1/19)

Our new tool, KA-Search, allows exhaustive yet efficient antibody mining through billions of natural sequences (from OAS and/or your private data), enabling new avenues in function & immunogenicity prediction. Work led by @HegelundOlsen and @brennanaba. biorxiv.org/cgi/content/sh…

In the works for a LONG while - a new clustering of antibody CDR structures to update North et al (2011)/PyIgClassify. Clusters now have high electron density support & at least 10 sequences. DBSCAN helped to remove noise points. @biorxiv_bioinfo @build_models @PDBeurope @PDBj_en

New work from our @UiO_LifeSci ImmunoLingo convergence environment in which we leverage linguistics to formalize the antibody sequence language. arxiv.org/abs/2209.12635. Work led by @MaiSpaceHa, @PRobertImmodels, @pandaisikit. See🧵below by @MaiSpaceHa for details.

We are very happy to share our review of #machinelearning methods in #antibodydiscovery #antibodydevelopment realized with Sandeep Kumar and Victor Greiff (@victorgreiff) has been published in Briefings in Bioinformatics naturalantibody.com/use-case/machi…

@WyattMcDonnell @naturalantibody @DeepMind Yes thx a lot, it was actually from here: nature.com/articles/s4200…

Can @DeepMind #Alphafold help in #antibodydiscovery? We tested its potential - find out what we discovered in this article naturalantibody.com/use-case/deepm…

Really excited to announce that AntiBERTa is now published in @Patterns_CP! Here we describe a transformer model that demonstrates understanding of antibody sequences 🧵 (1/6) #machinelearning #antibodies #drugdiscovery cell.com/patterns/fullt…

Excited to share IgFold, our new method for fast, accurate antibody structure prediction! IgFold achieves state-of-the-art accuracy for conventional paired antibodies and provides useful error estimates for its structures. biorxiv.org/content/10.110…

The evolutionary velocity paper ended on a cliffhanger: protein language models could predict evolution retrospectively, but could they also run evolution forward to prospectively design new proteins? So, I retrained as a protein biochemist to find out... biorxiv.org/content/10.110…

Very happy that our review on the "Progress and challenges for the machine learning-based design of fit-for-purpose monoclonal antibodies" is finally out. tandfonline.com/doi/full/10.10…

Happy to share our work on multi-dimensional comparison of immune repertoires. Amazing work by @cedricrweber. Great collaboration with Xiao Liu, @ReddyLab_ETHZ, @mkuijjer, and, as always, @SandveGeir. Main findings in 🧵. biorxiv.org/content/10.110….

The @alchemabtx team are excited by the huge growth of BCR repertoire sequencing and analysis! We collected a list of papers in this space – feel free to share and contribute👍 #BCRs #antibodies #papers #datasets #machinelearning github.com/alchemab/bcr-r…

Our work on generative language modeling for antibody sequences is now out on bioRxiv! 🧬 We take inspiration from text infilling and task our model with generating spans of amino acids within antibodies. 1/3 biorxiv.org/content/10.110…

The DB issue of NAR will have both INDI (db of nanobodies from structures, ncbi, patents, ngs, manual) academic.oup.com/nar/advance-ar… + nanobody structure update to sabdab (academic.oup.com/nar/advance-ar…). Good stuff @conctaylor @mijr12 @OPIGlets

Recent work on text mining vaccine-related news, quantifying volume & sentiment of coverage. @JemielniakD @ProfSamirBhatt medrxiv.org/content/10.110…

Preprint for AbDiver, tool to navigate the ngs/airr b-cell receptor diversity space for therapeutic applications : biorxiv.org/content/10.110…

Navigating millions of human #antibody sequences in next-generation sequencing #NGS repositories is hard, so scientists now turn to solutions like AbDiver for performing a fast natural reference for query sequences. Check out the use cases in our preprint: naturalantibody.com/use-case/abdiv…

@alchemabtx's tech team are delighted to present our latest pre-print on an antibody-specific language model, AntiBERTa, on bioRxiv! #antibodies #transformers #biotech (1/5) biorxiv.org/content/10.110…