GAMA Miguel Angel π¦ββ¬π
@miangoar
Followers
1K
Following
50K
Media
505
Statuses
4K
Biologist that navigates in the oceans of diversity through the space-time | MSc in Biochem/Bioinfo @ibt_unam π²π½ | Protein evo, metagenomics & AI/ML/DL
Mexico
Joined November 2020
RT @ProteinBoston: Have no fear if you weren't in the packed audience this week for @MartinPacesa's awesome seminar, because you can checkβ¦.
0
10
0
RT @DidiKieran: AtomWorks is out! Building upon @biotite_python, we built for a toolkit for all things biomolecules and trained RF3 with iβ¦.
0
24
0
Generate videos in just a few seconds. Try Grok Imagine, free for a limited time.
359
649
3K
tldr. They created AtomWorks, a high-level API (similar to Keras) for biomolecular stuff that simplifies data processing and training. Using it, they trained RF3 and Protein/Ligand-MPNN with improved chirality handling in just a few lines of code.
Accelerating Biomolecular Modeling with AtomWorks and RF3. π New preprint from David Baker!π. 1. A new framework called AtomWorks has been introduced to revolutionize biomolecular modeling. AtomWorks provides a unified and modular platform for developing state-of-the-art
1
23
110
RT @duguyuan: UpdateοΌπ The protein universe just got 20x bigger. π€―.Search ProTrek database of 6 billion proteins using text, sequence, or 3β¦.
search-protrek.com
0
14
0
RT @ScienceMagazine: Researchers have developed a #DeepLearning system called BioEmu that rapidly generates diverse protein conformations,β¦.
0
136
0
In the AF2-era would be:.- AFDB.- Geoformer.- ESMFold.- AFDB clustering (2 papers).- OpenFold.- The Encyclopedia of Domains.- AlphaFlow/Bioemu.- BindCraft. Surely Iβm forgetting several more (I only considered works that use the AF arch. or the AFDB).
0
0
2
Today the AlphaFold3 paper reached 1M access π₯³ Curiously AF2 took ~19 months to reach that numbers (July 2021 β March 2023), whereas AF3 took ~15 months (May 2024 β August 2025). IMO the most relevant works in the post AF3 era would be:.- AF3-reproductions .- BoltzDesign1.
Today the AlphaFold2 paper reached 2 million accesses according to Nature's statistics π₯³ As Janet Thornton said: βI think getting another AlphaFold moment like that will be very difficultβ. Anyway . congrats to the whole team behind AF2.
1
2
30
When AlphaFold predicts a protein structure with >95 pLDDT, you can say itβs accurate AF (as fuck) π€ͺ.
when chatgpt makes you laugh that's an llmao.
0
0
4
RT @CellCellPress: The new issue is outπ On the cover, Haziza et al. image high-frequency voltage dynamics in multiβ¦.
0
22
0
N = 59. It seems that proteinologist prefer AF2 over AF3 when it comes to monomers or cases that don't involve biomolecular interactions.
Iβve read opinions from colleagues who prefer AlphaFold2 predictions over those from AF3 due to the hallucinations in the latter. Iβm curious, which system do you prefer? π€.
0
0
5
For protein science/bioinformatics:.MMseqs.Diamond.FoldSeek.BioTite.ColabFold.BioPython.pdb-tools.Seqkit.hh-suite.IQ-Tree.MAFFT.TMAlign.PyRosetta (Just released in 2024).DSSP.BLAST.ProteinMPNN.RFDiff.AlphaFold2.ESM.BindCraft .RDKit.Boltz.HMMER.GROMACS. Which others?.
Huge thanks to all the open source projects that've made a lot of the tech we rely on in the world possible:. Linux.Git.FFmpeg .PyTorch & TensorFlow .Apache & Nginx .MySQL, PostgreSQL, SQLite.Chromium & Firefox.GCC & LLVM.Docker & Kubernetes. Also, all the open-weight LLMs. and.
10
90
548
0
15
0
RT @yoakiyama: Excited to share work with @ZhidianZ, Milot Mirdita, Martin Steinegger, and @sokrypton ..TLDR: We inβ¦.
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
Iβve read opinions from colleagues who prefer AlphaFold2 predictions over those from AF3 due to the hallucinations in the latter. Iβm curious, which system do you prefer? π€.
2
0
3
RT @ferruz_noelia: Iβve thoroughly enjoyed reading two (VERY!) recent papers that model protein sequences by retrieving evolutionary informβ¦.
0
28
0
The effect of making your research and trained models open source π₯
2
4
45