You're asked to process single-cell sequencing data, what should you do first? Before downloading the data, it is *crucial* to understand the structure of the reads. Where are the barcodes? UMIs? Genomics features?

1/4 last week, we saw an evaluation of SC foundation models in zero-shot settings; a new study tested fine-tuning and few-shot abilities in cell classification, and results are again similar; some perform on par or underperform vs logistics regression biorxiv.org/content/10.110…

# On the "hallucination problem" I always struggle a bit with I'm asked about the "hallucination problem" in LLMs. Because, in some sense, hallucination is all LLMs do. They are dream machines. We direct their dreams with prompts. The prompts start the dream, and based on the

Totally agree. A computational model truly shines when it builds on simple known biology rules. These known biological rules build trust into the algorithm and often prevent spurious or hallucinations.

How to pick a problem. Today in Cell. 1/53 cell.com/cell/fulltext/…

👏🏾This is a transformative reduction in long read (ONT) sequence error rate - changes the game. Congrats Mile!

Some fibroblasts are more equal than others.

Excited to share scPanel, a computational tool to identify sparse biomarker panels for patient stratification (think disease severity, treatment response) from #singlecell data. Developed by talented Yi Xie and co-supervised with Enrico Petretto. tinyurl.com/scpanelpaper

Very excited to present STAMP: An Interpretable spatially aware dimension reduction of spatial transcriptomics nature.com/articles/s4159…

Try ShinyCell2! Extended to support spatial, scATAC and multiomics data now!

ShinyCell2: An extended library for simple and sharable visualisation of spatial, peak-based and multi-omic single-cell data biorxiv.org/content/10.110…