Glad we’ve solved that, but that does not suffice to separate lasagne from mille-feuille. #OpenProblems

New OpenProblems paper out! 📝

Led by Malte Lücken with Smita Krishnaswamy, we present openproblems.bio – a community-driven platform benchmarking single-cell analysis methods.

Excited about transparent, evolving best practices for the field!

🔗 www.nature.com/articles/s41...

geneRNIB: a living benchmark for gene regulatory network inference www.biorxiv.org/content/10.1... 🧬🖥️🧪 github.com/openproblems...

About scRNAseq, they trained on a separate dataset and tested on OpenProblems benchmark. They picked 1 from 22 train-sets by manual inspection (no reasoning)🤔 Best result applies BBKNN to ComBat-corrected PCA embeddings rather than raw PCA. Could this “double normalization” risk overfitting?✨🧪 #AI4S

How to learn gene regulation patterns from multimodal single-cell data? To answer this question, we at OpenProblems organised the Kaggle competition on modality prediction. It ran back in 2022 but still remains the world's largest competition in the single-cell field. 1/17

Best models predictions are still far from ground truth, but we have anticipated this room for growth, as the platform is a living benchmark, where new methods can easily be integrated into the leaderboard via contributions on GitHub github.com/openproblems... . We're open to suggestions! 🧵7/8

scPRINT-2 is a cell foundation model trained on 350M cells from 16 organisms. It classifies cell types, denoises expression, predicts gene regulatory networks, and generates counterfactual profiles — all zero-shot.

75% accuracy on the OpenProblems live benchmark. scPRINT-1 was at 47%.
2/3

About scRNAseq, they trained on a separate dataset and tested on OpenProblems benchmark. They picked 1 from 22 train-sets by manual inspection (no reasoning)🤔 Best result applies BBKNN to ComBat-corrected PCA embeddings rather than raw PCA. Could this “double normalization” risk overfitting?

The beautiful aspect of the #SeymourConjectire and many #OpenProblems is that it seems intuitively true, but proving them has been difficult. Why do you think that is? #GraphTheory #Math #SSNC #MathPuzzles

Day 7 of the #SSNC facts! The #SeymourConjecture isn't the only conjecture in graph theory. There are related conjectures, like the #SullivanConjecture and the #DeanConjecture, that explore similar ideas about graph structure and neighborhoods. #GraphTheory #OpenProblems

Cracking Cell Complexity Together 💡

#OpenProblems is an open-source platform co-led by #HelmholtzMunich and Yale to benchmark and improve computational tools for single-cell genomics.

It offers standardized benchmarking, transparent evaluations, and cloud-based workflows.

7/
Open questions: How do partners find each other in the dark? What’s the drag vs. defense trade-off for mega-anemones? Are there shared genetic tweaks for building mineral layers? #EvoEco #OpenProblems

"OpenProblems built the commons for method evaluation; Seqera makes it scale; Data Intuitive ensures it’s robust. The result is a shared, reproducible backbone the entire community can run anywhere" - Robrecht Cannoodt (CTO, Data Intuitive)

"OpenProblems built the commons for method evaluation; Seqera makes it scale; Data Intuitive ensures it’s robust. The result is a shared, reproducible backbone the entire community can run anywhere" - Robrecht Cannoodt (CTO, Data Intuitive)

While related to the well-known #SeymourConjecture (proposed ~1990/95), the #SullivanConjecture is more recent (2006). Consequently, there's comparatively less research on this second neighborhood problem.
#GraphTheory #ComputerScience #OpenProblems #ResearchSkills #Math

Day 8 of SSNC facts! Yesterday we mentioned that the #SeymourConjecture isn't alone. Lets look at the #SullivanConjecture. It asks if the out-degrees of a node's second out-neighborhood are at least twice its in-degree. How are these conjectures similar? Different? #GraphTheory #OpenProblems #math

It's not just related conjectures! Tackling #OpenProblems like #SeymourConjecture reinforces classroom #Algorithms like #BFS and #DFS. Practical application deepens understanding. #GraphTheory #ComputerScience #Math #Dev #Programming

This is just a simple statement But it's also a decades-long puzzle? Is the #SeymourConjecture like an unbeatable video game challenge? What hidden complexity lies in these #OrientedGraphs? #MathProblems #Unsolved #GraphTheory #OpenProblems

💯 🙏 Also, I would like to acknowledge the important pioneering work from Geneformer, UCE, scFoundation and scGPT. Thanks to FlashAttention, pytorch, lightning, and scanpy for their toolkits. Thanks to Omnipath, Scenic+, Openproblems, Replogle et al. and Mc Calla et al.