Polaris

1/ Exciting update from the Multiomics Steering Committee on Polaris! Read the letter: polarishub.io/guidelines/m...

3/ Few highlights: — Many benchmarks reward models for finding what we already know — Drug discovery needs models that generalize, identify new biology, and flag problems early — Simple baselines still beat fancy models — We need clearer tasks, cleaner data, and shared frameworks

1/ Exciting update from the Multiomics Steering Committee on Polaris! Read the letter: polarishub.io/guidelines/m...

1/ 🚀 Our updated paper on practically significant ML method comparison is now published in JCIM!🚀 How do we ensure a model that tops a leaderboard also delivers a practical, real-world impact in drug-discovery? This paper offers a clear framework to address the field's replicability crisis.

3/ We're especially thrilled that this paper is open access thanks to the JCIM editor's choice program. We hope for the broad adoption of these guidelines to help close the gap between perceived progress and real-world impact in drug discovery.

1/ 🚀 Our updated paper on practically significant ML method comparison is now published in JCIM!🚀 How do we ensure a model that tops a leaderboard also delivers a practical, real-world impact in drug-discovery? This paper offers a clear framework to address the field's replicability crisis.

(1/3) 🎉 Exciting Updates: Antiviral Unblinded Datasets! 🎉 Following the release of the ASAP Discovery x OpenADMET antiviral challenge leaderboard, the evaluation code is now open-sourced and the unblinded datasets are available on Polaris. Unblinded Datasets: polarishub.io/datasets?acc...

(2/3) 📢 In addition, ASAP is hosting a virtual showcase tomorrow, April 30 at 11AM ET / 5PM BST featuring top performers from the challenge. Speakers will present their methods, background, and post-hoc analyses now that the test data is live. Register for the link: docs.google.com/forms/d/e/1F...

We're thrilled to announce the formation of the new Multiomics Steering Committee! By "multiomics" we refer to the integration of high-dimensional data modalities at the cellular level, encompassing molecular (DNA, RNA, protein) and phenotypic (imaging-based, multi-cell-line viability) readouts. 🧵

Led by experts from academia and industry, the committee includes: Oren Kraus (Recursion), Judith Mueller (Merck), Ali Denton (Valence Labs), Fabian J. Theis (Helmholtz Munich), Paula A. Marin Zapata (Bayer), Anne E. Carpenter (Broad Institute), Ian Barrett (AstraZeneca)... 🧵

We're thrilled to announce the formation of the new Multiomics Steering Committee! By "multiomics" we refer to the integration of high-dimensional data modalities at the cellular level, encompassing molecular (DNA, RNA, protein) and phenotypic (imaging-based, multi-cell-line viability) readouts. 🧵

We're thrilled to announce the formation of the new Multiomics Steering Committee! By "multiomics" we refer to the integration of high-dimensional data modalities at the cellular level, encompassing molecular (DNA, RNA, protein) and phenotypic (imaging-based, multi-cell-line viability) readouts. 🧵

and Rob Moccia (formerly Pfizer), Cuong Q. Nguyen (formerly GSK), Nate Robichaud (Nomic), William J. Godinez (Novartis), Tommaso Biancalani (Genentech), and Kristina Ulicna (Valence Labs). 🧵

🚀 Antiviral Challenge Results! 🚀 After weeks of competition, and nearly 400 submissions, the antiviral challenge has concluded! Thank you to all of our participants. Each submission came with a report — check out the full leaderboard and find more details 👇

📢 We value your feedback! Help us improve future challenges, share your thoughts. 📄 Feedback Form: share-eu1.hsforms.com/23XmDlumfQBu...

🚀 Antiviral Challenge Results! 🚀 After weeks of competition, and nearly 400 submissions, the antiviral challenge has concluded! Thank you to all of our participants. Each submission came with a report — check out the full leaderboard and find more details 👇

For detailed metrics, leaderboards, and further information, see the challenge page: antiviral.polarishub.io

🚀 Leaderboard Update: Antiviral Challenge! 🚀 The latest rankings are in—check out where you stand: antiviral.polarishub.io Missed this update? No worries, you still have time to compete! The challenge runs until March 10th at midnight. Get started today: polarishub.io/competitions

Today, we’re excited to welcome Nicola Richmond (@recursionpharma.bsky.social), Michael Reutlinger (Roche), and Nils Weskamp (@boehringerglobal.bsky.social) to the Small Molecule Steering Committee at Polaris! Learn more about what the steering committee is up to: polarishub.io/guidelines

⚡ The intermediate leaderboard for the antiviral challenge is now live! 👀 See how you placed: antiviral.polarishub.io The challenge officially ends on March 10th, but there's one last leaderboard update before the final results. Submit before midnight on Feb 26th! polarishub.io/competitions

You’ve heard us talk about our mission to be the source of truth for the ML community in drug discovery. But what does that really mean, and how does it shape the way Polaris is built? In the latest blog, we break down the key design choices that make this possible: polarishub.io/blog/reprodu...

🪑 Consistent benchmarks — we store the exact indices used in each split and each metric is codified in a single, transparent implementation. Explore the benchmarks: polarishub.io/benchmarks?a...

You’ve heard us talk about our mission to be the source of truth for the ML community in drug discovery. But what does that really mean, and how does it shape the way Polaris is built? In the latest blog, we break down the key design choices that make this possible: polarishub.io/blog/reprodu...

💽 Immutable datasets — once a dataset is uploaded to Polaris, it's permanently locked. Every time you or anyone else downloads the dataset, you'll get the same version. Explore datasets today: polarishub.io/datasets?acc...

💡Have questions about the data behind the antiviral challenge?💡 This week, @asapdiscovery.bsky.social and @omsf.io are hosting 3 virtual office hour sessions with the experimentalists behind the data! RSVP here: docs.google.com/forms/d/e/1F...

🗓️ Office Hours Schedule 🗓️ Jan 30th: 9 AM GMT / 10 AM CET Jan 30th: 8:30 AM PT / 11:30 AM ET Jan 31st: 8 AM PT / 11 AM ET docs.google.com/forms/d/e/1F...

MassSpecGym is the largest publicly available collection of mass spectra data with 231K spectra for 29K unique molecular structures. 33% of the dataset was generated from newly measured, in-house data. 🛡️The dataset is now certified on Polaris! polarishub.io/datasets/rom... youtu.be/G8ZnVRm0ogc

But decoding molecular structures from mass spectra is incredibly challenging, even for human experts. That’s why researchers are increasingly turning to machine learning to tackle this problem. ML aims to make structure elucidation from mass spec data faster, more accurate, and more scalable

MassSpecGym is the largest publicly available collection of mass spectra data with 231K spectra for 29K unique molecular structures. 33% of the dataset was generated from newly measured, in-house data. 🛡️The dataset is now certified on Polaris! polarishub.io/datasets/rom... youtu.be/G8ZnVRm0ogc

Mass spec plays a key role in identifying and characterizing molecules, particularly in biological and environmental samples. This is vital for advancing biomedical and chemical sciences—especially since ~half of today’s pill-based medicines were first discovered in nature.