Having fun with microbial genomic and machine learning @TelAvivUni
Israel
- Tweets 225
- Following 359
- Followers 641
- Likes 595
So many bacterial genomes are being edited… Could we spot them in the wild even without obvious markers??? Worry not! Our @EdanGabay has you covered. In our new preprint, she pinpoints such genes disrupting the natural genomic "grammar": biorxiv.org/content/10.64898… 1/4
3
6
31
1,412
Prefer just to run it from your browser? We've got you covered with our web server at app.famus.bursteinlab.org/. 7/8
1
1
102
Is it fast? We also built a lightweight version of FAMUS that runs close to the speed of a plain hmmsearch. Both the full and lightweight versions can run on CPU or (a bit faster) on GPU. 5/8
1
87
Benchmarking on KEGG and PANTHER, FAMUS improves on KofamScan and InterProScan when most sequences are unannotated, a realistic case for metagenomics and non-model organisms. This led us to compile FAMUS databases from KEGG, InterPro, eggNOG, and OrthoDB. 4/8
1
1
2
147
FAMUS uses supervised contrastive learning (SupCon) to explicitly separate known protein families from out-of-scope proteins. SupCon also handles sparsity, giving good classification even for small protein families. 2/8
1
272
⚡ Reduced alphabets yield shorter inputs and major runtime gains, while maintaining comparable, and sometimes improved, predictive performance. (4/5)
1
99
🔤 By combining Byte Pair Encoding (BPE) with reduced amino acid alphabets based on residue properties, we trained new pLMs and evaluated them across diverse biological tasks, like solubility, enzyme, PPI, and stability prediction. (3/5)
1
95
⚖️ Unlike natural languages, proteins aren’t clearly separated into “words”, making tokenization tricky. Short tokens create long sentences, while long tokens lead to a sparse vocabulary that is hard to learn. But reducing the alphabet size might help! (2/5)
1
1
1
190
2/4 B-PPI is a cross-attention model for bacterial PPI prediction at scale. Given protein pairs, it leverages ProstT5, a structure-aware protein language model, to generate embeddings, and outputs the interaction probability.
1
1
147
1/4 Ever wanted to predict bacterial protein-protein interactions (PPI) on a large scale?
We wanted to, but realized there’s no such algorithm that is both rapid and optimized for bacterial protein analysis.
This led our ⭐️Chen Agassy⭐️ to develop B-PPI: doi.org/10.64898/2025.12.23.…
1
11
29
2,055
5/6 💡 Discover how NLP is being applied to:
• Protein structure prediction 🏗️
• Taxonomic classification 🌳
• Mutational effect prediction 🔀
• Gene expression prediction 📈
And much more!
1
1
112
4/6 🧩 Tokenization challenges? We've got that covered too! Explore different approaches to breaking down biological sequences and their impact on model performance.
1
1
48
3/6 📚 We break down the evolution of NLP models in biology, from classic word2vec to cutting-edge transformers and hyena operators. Understand their strengths, limitations, and exciting applications!
1
1
59
2/6 🔬 We dive deep into how NLP techniques are revolutionizing the analysis of biological 'languages':
• DNA 🧬
• RNA 🧬
• Proteins 💪
• Entire genomes 🔍
Learn how these methods are unlocking new insights in genomics!
1
1
96
1/6 🧬📊 Curious about the buzz around NLP in biology? Feeling overwhelmed by the rapid developments? We've got you covered! Our review on NLP applications in genomics, by the wonderful @EllaRannon, is now out as a pre-print! #NLP #Bioinformatics arxiv.org/abs/2506.02212
2
11
31
2,026
💪Why so dramatic? DRAMMA detects antimicrobial resistance genes (ARGs) without relying on sequence similarity to known genes! 2/6
1
42
🚨New paper alert! The amazing @EllaRannon developed DRAMMA, a machine learning model that enhances antimicrobial resistance gene detection using diverse biological data🧬. Superbugs, beware! tinyurl.com/26b9x2ud
#Bioinformatics #MachineLearning #AMR 1/6
1
6
13
816
Promoters known to allow expression from ssDNA (Frpo) are widespread in the anti-defense islands. These may enable expression in the very early stages of conjugation, while plasmids are still in ssDNA form, leading to rapid and efficient evasion from host defense systems. 5/6
2
4
27
2,518
Anti-defense systems in the leading region tend to cluster into “islands”, which reside between defined boundaries.
These islands include different combinations of numerous anti-defense genes. 3/6
1
1
19
1,710
We examined an extensive set of conjugative elements from numerous genomes and metagenomes, and found that the leading region of plasmids, which is first to enter the recipient during conjugation, is a hotspot for a diverse repertoire of anti-defense genes. 2/6
1
20
1,711
Imagine you’re a plasmid conjugating into a recipient cell. How would you overcome CRISPR-Cas and other bacterial defense systems? Encoding anti-defense mechanisms is one obvious way.
But where? And how? @BruriaSamuel set out to explore! 🧵 1/6
biorxiv.org/content/10.1101/…
10
120
411
74,987