Gautam
@gautammalik.bsky.social
Research Assistant at University of Cambridge | Exploring deep learning in biology with big dreams of using AI to make drug discovery a little less complicated!🧬🖥️
Pinned
Gautam
@gautammalik.bsky.social
· Nov 22
Alright, let’s talk about 'Transforming docking with transformers!🔎' A few days ago, I came across Dockformer: A transformer-based molecular docking model (shoutout to @iwatobipen.bsky.social for sharing it). Here’s a quick breakdown of the architecture, so you can skip the deep dive!
Quick question for anyone doing transformer stuff in comp bio/chem or structured data!
Trying out masked modeling on a sparse setup, but not sure I'm going about it right. Curious how others have tackled this.
Trying out masked modeling on a sparse setup, but not sure I'm going about it right. Curious how others have tackled this.
June 14, 2025 at 5:22 AM
Quick question for anyone doing transformer stuff in comp bio/chem or structured data!
Trying out masked modeling on a sparse setup, but not sure I'm going about it right. Curious how others have tackled this.
Trying out masked modeling on a sparse setup, but not sure I'm going about it right. Curious how others have tackled this.
Reposted by Gautam
The surprising ineffectiveness of molecular dynamics coordinates for predicting bioactivity with machine learning
Checks whether MD-derived 3D information helps for bioactivity and target predictions over just static 3D information. Often no 3D info at all is best..
P: chemrxiv.org/engage/chemr...
Checks whether MD-derived 3D information helps for bioactivity and target predictions over just static 3D information. Often no 3D info at all is best..
P: chemrxiv.org/engage/chemr...
January 10, 2025 at 7:22 AM
The surprising ineffectiveness of molecular dynamics coordinates for predicting bioactivity with machine learning
Checks whether MD-derived 3D information helps for bioactivity and target predictions over just static 3D information. Often no 3D info at all is best..
P: chemrxiv.org/engage/chemr...
Checks whether MD-derived 3D information helps for bioactivity and target predictions over just static 3D information. Often no 3D info at all is best..
P: chemrxiv.org/engage/chemr...
A young researcher’s perspective on the #DiffDock discussion between @gcorso.bsky.social and @prof-ajay-jain.bsky.social:
Honestly, I’m feeling both thrilled and a little lost. As someone new to the field, I can’t help but reflect on what this means for the future of docking and AI/ML in science.
Honestly, I’m feeling both thrilled and a little lost. As someone new to the field, I can’t help but reflect on what this means for the future of docking and AI/ML in science.
December 9, 2024 at 4:57 AM
A young researcher’s perspective on the #DiffDock discussion between @gcorso.bsky.social and @prof-ajay-jain.bsky.social:
Honestly, I’m feeling both thrilled and a little lost. As someone new to the field, I can’t help but reflect on what this means for the future of docking and AI/ML in science.
Honestly, I’m feeling both thrilled and a little lost. As someone new to the field, I can’t help but reflect on what this means for the future of docking and AI/ML in science.
Reposted by Gautam
Impressive work by @franknoe.bsky.social and team! A pragmatic tour-de-force combining experimental and predicted protein structures, MD simulations and experimental stability data to sample conformational ensembles of proteins. Think AlphaFold, but capturing multiple free energy minima.
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 8, 2024 at 11:21 AM
Impressive work by @franknoe.bsky.social and team! A pragmatic tour-de-force combining experimental and predicted protein structures, MD simulations and experimental stability data to sample conformational ensembles of proteins. Think AlphaFold, but capturing multiple free energy minima.
Reposted by Gautam
When the size of test data is 5 compounds and accuracy is 100% 😌
December 6, 2024 at 9:15 PM
When the size of test data is 5 compounds and accuracy is 100% 😌
Reposted by Gautam
There's a new #RDKit blog post introducing some new functionality that I'm really excited about: doing efficient substructure and similarity searches in very large chemical libraries:
greglandrum.github.io/rdkit-blog/p...
#ChemSky
greglandrum.github.io/rdkit-blog/p...
#ChemSky
Introducing Synthon Searching – RDKit blog
Searching unreasonably large chemical spaces in reasonable amounts of time.
greglandrum.github.io
December 3, 2024 at 7:21 AM
There's a new #RDKit blog post introducing some new functionality that I'm really excited about: doing efficient substructure and similarity searches in very large chemical libraries:
greglandrum.github.io/rdkit-blog/p...
#ChemSky
greglandrum.github.io/rdkit-blog/p...
#ChemSky
Reposted by Gautam
#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.
Reposted by Gautam
It’s not real-world ready but a good foundation to explore. And yes, science does need a protein emoji!
github.com/gautammalik-...
github.com/gautammalik-...
GitHub - gautammalik-git/BindAxTransformer: BindAxTransformer is a transformer-based model trained on protein-ligand interactions using self-supervised learning. This repository provides a detailed im...
BindAxTransformer is a transformer-based model trained on protein-ligand interactions using self-supervised learning. This repository provides a detailed implementation and educational resource, sh...
github.com
November 22, 2024 at 7:46 PM
It’s not real-world ready but a good foundation to explore. And yes, science does need a protein emoji!
github.com/gautammalik-...
github.com/gautammalik-...
Reposted by Gautam
To wrap up, I’m curious about your thoughts on the future of docking models. Will the next breakthrough be GNN-based, transformer-based, or something like generative models (e.g., Diffusion)? I'd love to hear your opinions on what direction the field is heading. Let me know your thoughts!
November 22, 2024 at 7:46 PM
To wrap up, I’m curious about your thoughts on the future of docking models. Will the next breakthrough be GNN-based, transformer-based, or something like generative models (e.g., Diffusion)? I'd love to hear your opinions on what direction the field is heading. Let me know your thoughts!
Alright, let’s talk about 'Transforming docking with transformers!🔎' A few days ago, I came across Dockformer: A transformer-based molecular docking model (shoutout to @iwatobipen.bsky.social for sharing it). Here’s a quick breakdown of the architecture, so you can skip the deep dive!
November 22, 2024 at 7:46 PM
Alright, let’s talk about 'Transforming docking with transformers!🔎' A few days ago, I came across Dockformer: A transformer-based molecular docking model (shoutout to @iwatobipen.bsky.social for sharing it). Here’s a quick breakdown of the architecture, so you can skip the deep dive!
Reposted by Gautam
Check this out for using AI in daily life as a scientist!
Blog post on using AI in academia from @arjunraj.bsky.social and the great folks at Mid Career PI Slack!
Finally, a How To guide for how to use AI effectively in academia.
arjun-raj-lab.gitbook.io/arjun-rajs-t...
Finally, a How To guide for how to use AI effectively in academia.
arjun-raj-lab.gitbook.io/arjun-rajs-t...
Using AI in academia | Arjun Raj's Tools For Science
Arjun Raj, with input from Anne Carpenter and the members of Mid-Career PI Slack
arjun-raj-lab.gitbook.io
November 22, 2024 at 5:49 PM
Check this out for using AI in daily life as a scientist!
Reposted by Gautam
We added another 100 citations (total 577) in the latest version of our review on LLMs and agents in chemistry. Take a look!
arxiv.org/abs/2407.01603
arxiv.org/abs/2407.01603
November 18, 2024 at 2:05 AM
We added another 100 citations (total 577) in the latest version of our review on LLMs and agents in chemistry. Take a look!
arxiv.org/abs/2407.01603
arxiv.org/abs/2407.01603
Reposted by Gautam
Saw some old computer friends I’ve used, others I’ve read about, at the Computer History Museum tonight. And mice, lots of mice.
November 21, 2024 at 6:36 AM
Saw some old computer friends I’ve used, others I’ve read about, at the Computer History Museum tonight. And mice, lots of mice.
Reposted by Gautam
If you really love your protein of interest commit to a tattoo @ardemp.bsky.social
November 22, 2024 at 2:29 AM
If you really love your protein of interest commit to a tattoo @ardemp.bsky.social
Reposted by Gautam
Interesting new paper on large scale small molecule model pre-training for property prediction by Recursion - pretraining in this domain isn’t necessarily as easy to get to work as in inages or language rdcu.be/dZWa9
MolE: a foundation model for molecular graphs using disentangled attention
Nature Communications - Predictive models for chemistry are typically trained on small data sets, making it difficult to generalize well. Here, the authors describe a foundation model trained on...
rdcu.be
November 12, 2024 at 9:49 PM
Interesting new paper on large scale small molecule model pre-training for property prediction by Recursion - pretraining in this domain isn’t necessarily as easy to get to work as in inages or language rdcu.be/dZWa9
Reposted by Gautam
November 22, 2024 at 2:04 AM
Reposted by Gautam
Reposted by Gautam
Since this is my first post here, I figured sharing my work would be a great way to make a splash. 🚀 I’ve put together two GitHub repositories for AutoDock GPU and CPU docking, with step-by-step guides perfect for those new to docking and simulations.
github.com/gautammalik-...
github.com/gautammalik-...
GitHub - gautammalik-git/AutoDock-GPU-Pipeline: This pipeline facilitates setting up ligand docking against a protein using AutoDock-GPU. It streamlines the process of docking a ligand library onto a ...
This pipeline facilitates setting up ligand docking against a protein using AutoDock-GPU. It streamlines the process of docking a ligand library onto a protein structure, leveraging the enhanced pe...
github.com
November 17, 2024 at 6:57 AM
Since this is my first post here, I figured sharing my work would be a great way to make a splash. 🚀 I’ve put together two GitHub repositories for AutoDock GPU and CPU docking, with step-by-step guides perfect for those new to docking and simulations.
github.com/gautammalik-...
github.com/gautammalik-...
Reposted by Gautam
Here's a Cheminformatics Starter
pack! Let me know if you would like to be added! And more importantly, do encourage people to join the blue skies :D
pack! Let me know if you would like to be added! And more importantly, do encourage people to join the blue skies :D
November 17, 2024 at 4:43 AM
Here's a Cheminformatics Starter
pack! Let me know if you would like to be added! And more importantly, do encourage people to join the blue skies :D
pack! Let me know if you would like to be added! And more importantly, do encourage people to join the blue skies :D