Kyle Tretina
@allthingsapx.bsky.social
Product Marketing Lead @NVIDIA
| PhD @UMBaltimore | omics, immuno/micro, AI/ML | 🇺🇸🇸🇰 |
Posts are my own views, not those of my employer.
| PhD @UMBaltimore | omics, immuno/micro, AI/ML | 🇺🇸🇸🇰 |
Posts are my own views, not those of my employer.
Reposted by Kyle Tretina
@allthingsapx.bsky.social on the new Biohub initiative by the Zucks
November 9, 2025 at 8:57 AM
@allthingsapx.bsky.social on the new Biohub initiative by the Zucks
Reposted by Kyle Tretina
MMseqs2-GPU sets new standards in single query search speed, allows near instant search of big databases, scales to multiple GPUs and is fast beyond VRAM. It enables ColabFold MSA generation in seconds and sub-second Foldseek search against AFDB50. 1/n
📄 www.nature.com/articles/s41...
💿 mmseqs.com
📄 www.nature.com/articles/s41...
💿 mmseqs.com
GPU-accelerated homology search with MMseqs2 - Nature Methods
Graphics processing unit-accelerated MMseqs2 offers tremendous speedups for homology retrieval from metagenomic databases, query-centered multiple sequence alignment generation for structure predictio...
www.nature.com
September 21, 2025 at 8:06 AM
MMseqs2-GPU sets new standards in single query search speed, allows near instant search of big databases, scales to multiple GPUs and is fast beyond VRAM. It enables ColabFold MSA generation in seconds and sub-second Foldseek search against AFDB50. 1/n
📄 www.nature.com/articles/s41...
💿 mmseqs.com
📄 www.nature.com/articles/s41...
💿 mmseqs.com
Near-real-time protein structures change science:
It means:
→ Next-gen protein AI data waves
→ Interactive protein design loops (DMTA in hours)
→ Proteome-scale insights with fewer resources
It means the bottleneck doesn't have to be compute.
It's close (preprint below).
It means:
→ Next-gen protein AI data waves
→ Interactive protein design loops (DMTA in hours)
→ Proteome-scale insights with fewer resources
It means the bottleneck doesn't have to be compute.
It's close (preprint below).
September 16, 2025 at 1:59 PM
Near-real-time protein structures change science:
It means:
→ Next-gen protein AI data waves
→ Interactive protein design loops (DMTA in hours)
→ Proteome-scale insights with fewer resources
It means the bottleneck doesn't have to be compute.
It's close (preprint below).
It means:
→ Next-gen protein AI data waves
→ Interactive protein design loops (DMTA in hours)
→ Proteome-scale insights with fewer resources
It means the bottleneck doesn't have to be compute.
It's close (preprint below).
Does anyone here care about biomolecular AI? Who should I follow?
August 30, 2025 at 2:07 AM
Does anyone here care about biomolecular AI? Who should I follow?
🧬 Introducing La‑Proteina:
a partially‑latent flow‑matching model that co‑generates sequence + all‑atom structure for proteins up to 800 aa 🧬
Side‑chains live in latents, backbone explicit → 75 % codesign & SOTA motif scaffolds 🔥
a partially‑latent flow‑matching model that co‑generates sequence + all‑atom structure for proteins up to 800 aa 🧬
Side‑chains live in latents, backbone explicit → 75 % codesign & SOTA motif scaffolds 🔥
July 15, 2025 at 6:26 PM
🧬 Introducing La‑Proteina:
a partially‑latent flow‑matching model that co‑generates sequence + all‑atom structure for proteins up to 800 aa 🧬
Side‑chains live in latents, backbone explicit → 75 % codesign & SOTA motif scaffolds 🔥
a partially‑latent flow‑matching model that co‑generates sequence + all‑atom structure for proteins up to 800 aa 🧬
Side‑chains live in latents, backbone explicit → 75 % codesign & SOTA motif scaffolds 🔥
July 15, 2025 at 6:25 PM
Boltz-2 just dropped: open-source AI that predicts both protein complex folds ✚ binding affinities in one shot 🚀
This is a win for protein AI, but let's not forget MSAs, the bioinformatics backbone many structure models lean on.
This is a win for protein AI, but let's not forget MSAs, the bioinformatics backbone many structure models lean on.
June 13, 2025 at 2:04 PM
Boltz-2 just dropped: open-source AI that predicts both protein complex folds ✚ binding affinities in one shot 🚀
This is a win for protein AI, but let's not forget MSAs, the bioinformatics backbone many structure models lean on.
This is a win for protein AI, but let's not forget MSAs, the bioinformatics backbone many structure models lean on.
MMseqs2-GPU is available as a downloadable NVIDIA NIM microservice (MSA-Search)!
June 13, 2025 at 1:56 PM
MMseqs2-GPU is available as a downloadable NVIDIA NIM microservice (MSA-Search)!
Reposted by Kyle Tretina
📢📢 "Proteina: Scaling Flow-based Protein Structure Generative Models"
#ICLR2025 (Oral Presentation)
🔥 Project page: research.nvidia.com/labs/genair/...
📜 Paper: arxiv.org/abs/2503.00710
🛠️ Code and weights: github.com/NVIDIA-Digit...
🧵Details in thread...
(1/n)
#ICLR2025 (Oral Presentation)
🔥 Project page: research.nvidia.com/labs/genair/...
📜 Paper: arxiv.org/abs/2503.00710
🛠️ Code and weights: github.com/NVIDIA-Digit...
🧵Details in thread...
(1/n)
March 4, 2025 at 5:09 PM
📢📢 "Proteina: Scaling Flow-based Protein Structure Generative Models"
#ICLR2025 (Oral Presentation)
🔥 Project page: research.nvidia.com/labs/genair/...
📜 Paper: arxiv.org/abs/2503.00710
🛠️ Code and weights: github.com/NVIDIA-Digit...
🧵Details in thread...
(1/n)
#ICLR2025 (Oral Presentation)
🔥 Project page: research.nvidia.com/labs/genair/...
📜 Paper: arxiv.org/abs/2503.00710
🛠️ Code and weights: github.com/NVIDIA-Digit...
🧵Details in thread...
(1/n)
I’m @neurips24! Let’s chat😁
December 10, 2024 at 6:28 PM
I’m @neurips24! Let’s chat😁
Reposted by Kyle Tretina
"Are there at least 3 simulations per simulation condition with statistical analysis?"
From @commsbio.bsky.social's "Reliability and reproducibility checklist for molecular dynamics simulations" (doi.org/10.1038/s420...)
IMO the number 3 is meaningless and could equally well be 1 or 1000
From @commsbio.bsky.social's "Reliability and reproducibility checklist for molecular dynamics simulations" (doi.org/10.1038/s420...)
IMO the number 3 is meaningless and could equally well be 1 or 1000
December 7, 2024 at 6:33 PM
"Are there at least 3 simulations per simulation condition with statistical analysis?"
From @commsbio.bsky.social's "Reliability and reproducibility checklist for molecular dynamics simulations" (doi.org/10.1038/s420...)
IMO the number 3 is meaningless and could equally well be 1 or 1000
From @commsbio.bsky.social's "Reliability and reproducibility checklist for molecular dynamics simulations" (doi.org/10.1038/s420...)
IMO the number 3 is meaningless and could equally well be 1 or 1000
Reposted by Kyle Tretina
Super excited to preprint our work on developing a Biomolecular Emulator (BioEmu): Scalable emulation of protein equilibrium ensembles with generative deep learning from @msftresearch.bsky.social ch AI for Science.
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
December 6, 2024 at 8:39 AM
Super excited to preprint our work on developing a Biomolecular Emulator (BioEmu): Scalable emulation of protein equilibrium ensembles with generative deep learning from @msftresearch.bsky.social ch AI for Science.
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
DiffDock was the first time a traditional drug discovery simulation task was represented as a generative AI task AFAIK.
Recent DiffDock versions + other DL models are advancing rapidly + solving real problems for researchers.
Let's have a balanced conversation about it.
arxiv.org/abs/2412.02889
Recent DiffDock versions + other DL models are advancing rapidly + solving real problems for researchers.
Let's have a balanced conversation about it.
arxiv.org/abs/2412.02889
Deep-Learning Based Docking Methods: Fair Comparisons to Conventional Docking Workflows
The diffusion learning method, DiffDock, for docking small-molecule ligands into protein binding sites was recently introduced. Results included comparisons to more conventional docking approaches, wi...
arxiv.org
December 6, 2024 at 3:39 PM
DiffDock was the first time a traditional drug discovery simulation task was represented as a generative AI task AFAIK.
Recent DiffDock versions + other DL models are advancing rapidly + solving real problems for researchers.
Let's have a balanced conversation about it.
arxiv.org/abs/2412.02889
Recent DiffDock versions + other DL models are advancing rapidly + solving real problems for researchers.
Let's have a balanced conversation about it.
arxiv.org/abs/2412.02889
Reposted by Kyle Tretina
#CASP16 results are in! Template-based VFold seems to be lead method for nucleic acid structure prediction! AlphaFold2 and 3 still seem to be best methods for protein monomer and complex prediction.
November 30, 2024 at 10:28 PM
#CASP16 results are in! Template-based VFold seems to be lead method for nucleic acid structure prediction! AlphaFold2 and 3 still seem to be best methods for protein monomer and complex prediction.
If you're building AI models for drug discovery, you should check out the newly open-source #BioNeMo Framework:
code: github.com/NVIDIA/bione...
paper: arxiv.org/abs/2411.10548
docs: docs.nvidia.com/bionemo-fram...
explainer: t.co/7MOamSChGN
code: github.com/NVIDIA/bione...
paper: arxiv.org/abs/2411.10548
docs: docs.nvidia.com/bionemo-fram...
explainer: t.co/7MOamSChGN
GitHub - NVIDIA/bionemo-framework: BioNeMo Framework: For building and adapting AI models in drug discovery at scale
BioNeMo Framework: For building and adapting AI models in drug discovery at scale - NVIDIA/bionemo-framework
github.com
November 19, 2024 at 2:29 PM
If you're building AI models for drug discovery, you should check out the newly open-source #BioNeMo Framework:
code: github.com/NVIDIA/bione...
paper: arxiv.org/abs/2411.10548
docs: docs.nvidia.com/bionemo-fram...
explainer: t.co/7MOamSChGN
code: github.com/NVIDIA/bione...
paper: arxiv.org/abs/2411.10548
docs: docs.nvidia.com/bionemo-fram...
explainer: t.co/7MOamSChGN
Friends,
Real-time, accurate protein structure prediction has never felt so imminent.
Code: github.com/soedinglab/m...
Publication: www.biorxiv.org/content/10.1...
Blog: developer.nvidia.com/blog/boost-a...
Press: blogs.nvidia.com/blog/japan-s...
Real-time, accurate protein structure prediction has never felt so imminent.
Code: github.com/soedinglab/m...
Publication: www.biorxiv.org/content/10.1...
Blog: developer.nvidia.com/blog/boost-a...
Press: blogs.nvidia.com/blog/japan-s...
November 16, 2024 at 2:23 AM
Friends,
Real-time, accurate protein structure prediction has never felt so imminent.
Code: github.com/soedinglab/m...
Publication: www.biorxiv.org/content/10.1...
Blog: developer.nvidia.com/blog/boost-a...
Press: blogs.nvidia.com/blog/japan-s...
Real-time, accurate protein structure prediction has never felt so imminent.
Code: github.com/soedinglab/m...
Publication: www.biorxiv.org/content/10.1...
Blog: developer.nvidia.com/blog/boost-a...
Press: blogs.nvidia.com/blog/japan-s...
Reposted by Kyle Tretina
developer.nvidia.com/blog/boost-a...
Boost Alphafold2 Protein Structure Prediction with GPU-Accelerated MMseqs2
Nice improvements in speed
Boost Alphafold2 Protein Structure Prediction with GPU-Accelerated MMseqs2
Nice improvements in speed
Boost Alphafold2 Protein Structure Prediction with GPU-Accelerated MMseqs2 | NVIDIA Technical Blog
The ability to compare the sequences of multiple related proteins is a foundational task for many life science researchers. This is often done in the form of a multiple sequence alignment (MSA)…
developer.nvidia.com
November 14, 2024 at 6:10 AM
developer.nvidia.com/blog/boost-a...
Boost Alphafold2 Protein Structure Prediction with GPU-Accelerated MMseqs2
Nice improvements in speed
Boost Alphafold2 Protein Structure Prediction with GPU-Accelerated MMseqs2
Nice improvements in speed
Now available – DiffDock 2.0 NIM
This latest #NVIDIANIM update offers computational chemists and researchers a significant boost with 16% improved accuracy in identifying potential protein-small molecule interactions with
Test for free:
build.nvidia.com/mit/diffdock...
This latest #NVIDIANIM update offers computational chemists and researchers a significant boost with 16% improved accuracy in identifying potential protein-small molecule interactions with
Test for free:
build.nvidia.com/mit/diffdock...
diffdock model by mit | NVIDIA NIM
Predicts the 3D structure of how a molecule interacts with a protein.
build.nvidia.com
November 13, 2024 at 10:52 PM
Now available – DiffDock 2.0 NIM
This latest #NVIDIANIM update offers computational chemists and researchers a significant boost with 16% improved accuracy in identifying potential protein-small molecule interactions with
Test for free:
build.nvidia.com/mit/diffdock...
This latest #NVIDIANIM update offers computational chemists and researchers a significant boost with 16% improved accuracy in identifying potential protein-small molecule interactions with
Test for free:
build.nvidia.com/mit/diffdock...
Now available – an updated RFdiffusion NIM
This #NVIDIANIM enables researchers to efficiently design protein therapeutic candidates 1.9x faster due to accelerations in the inference engine, making their preclinical research smarter and less expensive.
Test for free:
build.nvidia.com/ipd/rfdiffus...
This #NVIDIANIM enables researchers to efficiently design protein therapeutic candidates 1.9x faster due to accelerations in the inference engine, making their preclinical research smarter and less expensive.
Test for free:
build.nvidia.com/ipd/rfdiffus...
rfdiffusion model by ipd | NVIDIA NIM
A generative model of protein backbones for protein binder design.
build.nvidia.com
November 13, 2024 at 10:52 PM
Now available – an updated RFdiffusion NIM
This #NVIDIANIM enables researchers to efficiently design protein therapeutic candidates 1.9x faster due to accelerations in the inference engine, making their preclinical research smarter and less expensive.
Test for free:
build.nvidia.com/ipd/rfdiffus...
This #NVIDIANIM enables researchers to efficiently design protein therapeutic candidates 1.9x faster due to accelerations in the inference engine, making their preclinical research smarter and less expensive.
Test for free:
build.nvidia.com/ipd/rfdiffus...
Accelerated computing is powering a new era in protein structure prediction, now with speed-of-light multiple sequence alignments.
#MMseqs2GPU now makes #AlphaFold2 predictions faster and more efficient than ever.
1/🧵
developer.nvidia.com/blog/boost-a...
#MMseqs2GPU now makes #AlphaFold2 predictions faster and more efficient than ever.
1/🧵
developer.nvidia.com/blog/boost-a...
Boost Alphafold2 Protein Structure Prediction with GPU-Accelerated MMseqs2 | NVIDIA Technical Blog
The ability to compare the sequences of multiple related proteins is a foundational task for many life science researchers. This is often done in the form of a multiple sequence alignment (MSA)…
developer.nvidia.com
November 13, 2024 at 10:03 PM
Accelerated computing is powering a new era in protein structure prediction, now with speed-of-light multiple sequence alignments.
#MMseqs2GPU now makes #AlphaFold2 predictions faster and more efficient than ever.
1/🧵
developer.nvidia.com/blog/boost-a...
#MMseqs2GPU now makes #AlphaFold2 predictions faster and more efficient than ever.
1/🧵
developer.nvidia.com/blog/boost-a...
Accelerated computing is revolutionizing protein structure prediction, starting with lightning-fast multiple sequence alignments.
Read more about #AlphaFold and #MMseqs2GPU:
developer.nvidia.com/blog/boost-a...
Read more about #AlphaFold and #MMseqs2GPU:
developer.nvidia.com/blog/boost-a...
Boost Alphafold2 Protein Structure Prediction with GPU-Accelerated MMseqs2 | NVIDIA Technical Blog
The ability to compare the sequences of multiple related proteins is a foundational task for many life science researchers. This is often done in the form of a multiple sequence alignment (MSA)…
developer.nvidia.com
November 13, 2024 at 9:32 PM
Accelerated computing is revolutionizing protein structure prediction, starting with lightning-fast multiple sequence alignments.
Read more about #AlphaFold and #MMseqs2GPU:
developer.nvidia.com/blog/boost-a...
Read more about #AlphaFold and #MMseqs2GPU:
developer.nvidia.com/blog/boost-a...
Analyzing extensive scientific literature and real-world evidence, Muse delivers comprehensive research and identifies optimal patient profiles and recruitment strategies, including materials tailored for diverse populations.
formation.bio/blog/introdu...
formation.bio/blog/introdu...
Introducing Muse
Formation Bio collaborates with Sanofi and OpenAI to Introduce Muse, a first of its kind AI tool to accelerate patient recruitment in drug development.
formation.bio
November 13, 2024 at 2:28 PM
Analyzing extensive scientific literature and real-world evidence, Muse delivers comprehensive research and identifies optimal patient profiles and recruitment strategies, including materials tailored for diverse populations.
formation.bio/blog/introdu...
formation.bio/blog/introdu...
I can't help but think TDC-2 is a peek into the future of therapeutic AI, where multimodal data integration and single-cell precision drive discovery.
biorxiv.org/content/10.1...
biorxiv.org/content/10.1...
biorxiv.org
November 13, 2024 at 2:24 PM
I can't help but think TDC-2 is a peek into the future of therapeutic AI, where multimodal data integration and single-cell precision drive discovery.
biorxiv.org/content/10.1...
biorxiv.org/content/10.1...