Garyk Brixi
@garykbrixi
Followers
449
Following
1K
Media
6
Statuses
147
PhD student at Stanford Genetics. DNA BERTologist
Joined November 2022
RT @BrianHie: It was great chatting with @NikoMcCarty in this Q&A about the past and future of AI in biology. Evolution is a unifying theo….
arcinstitute.org
As a pioneer at the interface of machine learning and biology, Brian Hie (X: @brianhie) is known especially for developing deep learning models that can interpret the language of entire genomes—not...
0
27
0
Protein language models are returning to their roots by incorporating homologs (Dayhoff, MSA Pairformer, Protriever, PoET), while some AlphaFold-inspired design models instead try avoid MSA addiction. The future may lie in adaptivity to available data; constrained or creative.
2
2
52
RT @yoakiyama: Excited to share work with @ZhidianZ, Milot Mirdita, Martin Steinegger, and @sokrypton ..TLDR: We in….
biorxiv.org
Recent efforts in protein language modeling have focused on scaling single-sequence models and their training data, requiring vast compute resources that limit accessibility. Although models that use...
0
51
0
Evolution is so "overfit", reusing existing gene fragments. We want models that understand real biology and also generalize beyond nature's explored space. More work is needed to understand the tension and better define what overfitting means for different settings in bio.
We loved the discussion online spurred by @btnaughton on sequence composition of generated proteins. How much are these models simply "cheating" vs properly "simulating" nature? I think this a rich area for future research and has analogous tensions in text and NLP. To
0
3
58
Use test set perplexity and FPD to choose 'the best' model? Filter by pLDDT and scPerplexity? Maybe not!. One important takeaway is the need for better in-silico metrics both through prospective experiments and retrospective analysis of available data.
0
0
4
New synthetic and metagenomic data boosted experimental success while popular metrics failed to predict it. Read Ava's thread on the really cool models, analysis, and data resources!.
increasing model and data scale increased the fraction of proteins expressed by E. coli, and the highest expression success rate came from augmenting w/ structure-based synthetic data. data quality + diversity bring real gains in real-world protein expression!
1
6
24
RT @KevinKaichuang: In 1965, Margaret Dayhoff published the Atlas of Protein Sequence and Structure, which collated the 65 proteins whose a….
0
87
0
Evo 2 update: new dependency versions (torch, transformer engine, flash attn) and a docker option mean it should be easy to setup without needing to compile locally. Happy ATGC-ing!.
github.com
Genome modeling and design across all domains of life - ArcInstitute/evo2
2
21
131
RT @jkpritch: Staff scientist position (computational):. I am looking for a computational scientist to join my genomics lab at Stanford. Th….
0
32
0
RT @pdhsu: Delighted to announce @arcinstitute's Virtual Cell Challenge - a recurring, open, community-driven challenge to benchmark cellul….
0
65
0
RT @s6juncheng: Excited to share #AlphaGenome, a start of our AlphaGenome named journey to decipher the regulatory genome! The model matche….
0
210
0
RT @ras_nielsen: Got several responses exploring which software can deal with this. My point was that students should not rely on the preci….
0
7
0
RT @ruben_weitzman: 🚨ICML Paper Alert🚨.What if finding the right protein homologs wasn't a slow search, but a learned part of the model its….
0
21
0
RT @NotinPascal: 🚨 New paper 🚨 RNA modeling just got its own Gym! 🏋️ Introducing RNAGym, large-scale benchmarks for RNA fitness and structu….
0
49
0
RT @ChoYehlin: 🚀 Excited to release BoltzDesign1!. ✨ Now with LogMD-based trajectory visualization. 🔗 Demo: Feedbac….
0
72
0
RT @BiologyAIDaily: From Likelihood to Fitness: Improving Variant Effect Prediction in Protein and Genome Language Models. １．This study int….
0
19
0
RT @jxmnop: excited to finally share on arxiv what we've known for a while now:. All Embedding Models Learn The Same Thing. embeddings fro….
0
623
0
RT @HannesStaerk: Reading group tomorrow: @json_yim and @woodyahern present "Atom level enzyme active site scaffolding using RFdiffusion2"….
0
20
0
RT @nooryoussef03: 🚨 New in @ImmunityCP !.EVE-Vax, an AI model that anticipates future viral evolution and designs antigens to proactively….
0
22
0