HBKgenomics

Somewhere between bioinformatician and genomic researcher. Not good at executing routine works. Daydreaming coffee addict.

Art as research, Research as art

[Sergey Ovchinnikov @sokrypton]

Upper-left - academia: still struggling to raise $100K to support the project (grants, fellowships rejected). 😅

Upper-right - industry: raised $142 million in seed funding (company sold for much much more). 🙃

[HBKgenomics @HKgenomics]

So grateful to my collaborators Rachel Seongeun Kim, Milot Mirdita, Jaewon Yoon, and my advisor Martin Steinegger, plus the whole SteineggerLab. 🙏
🔗 https://t.co/536kvmW9xU
🧵7/7

[HBKgenomics @HKgenomics]

Does your designed active site already exist in nature? Is an uncharacterized protein hiding a catalytic site or a pocket? Folddisco can answer both, searching millions of structures for a 3D motif in seconds.
@NatureBiotech🧬
📄 https://t.co/gwRm8fSbIi
🧵1/7👇

[HBKgenomics @HKgenomics]

Webserver now pairs with Foldseek & FoldMason: search, motif detection & structure alignment in one workflow. Free, open source, prebuilt indices for AFDB50, PDB, ESM30, BFVD & more. With Milot Mirdita, Rachel Seongeun Kim & Cameron L. M. Gilchrist https://t.co/hI3ZjmD73O
🧵6/7

[Nature Biotechnology @NatureBiotech]

Structural motif search across the protein universe with Folddisco https://t.co/07rjoDIky9

[Jin Seok (Andy) Lee @iam_js_lee]

Structural variants are an untapped source of neoantigens but their messy representation holds us back. Enter Occam's Variant Grammar: a unified variant representation for long-read guided neoantigen discovery in personalized immunotherapy development.

https://t.co/f19UBkSGTR

[Biology+AI Daily @BiologyAIDaily]

Structural motif search across the protein-universe with Folddisco

1. A new computational tool, Folddisco, has been developed to tackle the computationally intensive problem of identifying similar protein structural motifs in vast collections of protein structures. It offers significant advancements in speed and efficiency.

2. Folddisco achieves its remarkable performance by utilizing an innovative index of position-independent geometric features, which includes side-chain orientation, and a unique rarity-based scoring system. This approach allows for rapid detection of both discontinuous and segment motifs.

3. This method is exceptionally fast and storage-efficient. It can index 53 million AlphaFold Database (AFDB50) structures into just 1.45 terabytes within 24 hours, proving to be over 18 times faster and requiring less than 3.5 times the storage of previous state-of-the-art methods.

4. A key innovation of Folddisco is its versatility, supporting searches for short motifs like zinc fingers, as well as complex, discontinuous segments, all within a single unified framework. This addresses a significant limitation of prior tools.

5. The tool enhances its accuracy by incorporating two additional torsion angle features that capture side-chain orientations, which are critical for recognizing enzyme-activity related motifs. A rarity-based scoring system further refines results by prioritizing specific and rare feature sets.

6. Folddisco has demonstrated superior accuracy and generalizability in benchmarks, outperforming existing methods like RCSB and pyScoMotif in detecting various protein motifs. It offers applications in functional annotation, identification of protein conformational states, and the detection of protein-protein interaction interfaces.

7. A webserver is available, enabling users to easily perform motif searches on major databases such as AFDB50 and PDB100, providing interactive structure visualizations and rapid results.

💻Code: https://t.co/C30tYEHlyw
📜Paper: https://t.co/gvPLIvwqHq
#ComputationalBiology #ProteinStructure #Bioinformatics #StructuralBiology #Folddisco

[Kyle Tretina, Ph.D. @AllThingsApx]

🧬Motifs are the new BLAST

Folddisco scans 53M AlphaFold structures for catalytic, binding or allosteric patterns in seconds🔎

18× quicker, 3.5× smaller, hits in seconds🌌

[HBKgenomics @HKgenomics]

This work was made possible by the incredible efforts of my lab @eunbelivable, Milot Mirdita and Martin Steinegger. 9/9

[HBKgenomics @HKgenomics]

Folddisco detects (partial) motifs, allowing for angle-length variations, by utilizing an index storing all residue pairs within 20Å encoded as geometric features. For efficiency, it omits positions and compresses ids through run-length encoding (1.4TB for 53M structures) 2/9

[HBKgenomics @HKgenomics]

We provide a user-friendly Folddisco webserver, enabling instant structural motif searches in PDB, AFDB-Proteomes, AFDB50, and ESMatlas (ESM30). Explore it here: https://t.co/hI3ZjmD73O https://t.co/yd1Vu2dtX1 8/9

[HBKgenomics @HKgenomics]

Folddisco finds similar (dis)continuous 3D motifs in large protein structure databases. Its efficient index enables fast active site annotation, protein conformational state analysis and PPI interface comparison. 1/9🧶🧬
📄 https://t.co/KRhRnZlcnr
🌐 https://t.co/hI3ZjmD73O