5/ An R package for FLASH-MM, including example case scenarios, is publicly available at:.📦

4/ FLASH-MM controls false positive rates and maintains high statistical power in differential expression analysis. We validate FLASH-MM using extensive simulations and real-world applications such as the tuberculosis immune atlas (~500K cells).

3/ FLASH-MM introduces a new algorithmic formulation of LMMs where computation is independent of the number of cells, delivering orders of magnitude improvement in speed and memory usage over standard approaches while maintaining accuracy.

2/ Differential expression analysis at the cell level requires accounting for hierarchical structure (e.g., cells nested within donors). Linear mixed models (LMMs) offer a principled framework but have been computationally prohibitive for large single-cell datasets - until now!.

🧵1/ Excited to share our new preprint: FLASH-MM - a fast and scalable method for single-cell differential expression using linear mixed models, enabling DE analysis in massive scRNA-seq datasets. 🔗#Changjiang_Xu @BioinVeronique @hheydarii @garybader1

sciRED is now officially out @NatureComms !🚀 The peer-reviewed version includes several new analyses beyond our preprint, so check it out even if you’ve seen it before! Huge thanks to @garybader1 and @talandrews for their incredible guidance.

RT @StephIrisPro: So excited to share the first paper from the lab!.We identified CD34 as a marker of the heart’s SAN pacemaker cells enabl….

RT @const_ae: There's a lot of excitement about foundation models and their ability to learn biology 🧬💻. But current tools for perturbation….

RT @DonnellyCentre: Congratulations to our 2024 Dorrington Award Winners! 🥳. Graduate students Heta Lad, Delaram Pouyabahar and Justin Sing….

RT @razoralign: sciRED: Interpretable single-cell factor decomposition

6/🔗 Read our full manuscript to dive deeper into sciRED's capabilities and our findings: @garybader1 @talandrews.

5/🐍 sciRED is available as a Python package! Find the installation instructions and tutorials on our GitHub:

4/ ➔ Guided clustering-free cell type annotation.➔ Discovered hidden biology such as rare cell type signatures missed in the original study.

3 / We applied sciRED to multiple scRNA-seq datasets. Here are some highlights:.➔ Identified general and cell-type specific covariate-related variation, such as sex-specific variations in a kidney map.➔ Reduced ambient RNA contamination.

2/ sciRED identifies biological signals in scRNAseq data, overcoming noise and technical factors. It removes confounding effects, improves interpretability through rotations, maps factors to known covariates, and identifies genes and processes these factors represent.

1/🚀Excited to share our latest work on sciRED, a new computational method for enhancing factor identification and interpretation in scRNA-seq data. sciRED extracts factors from gene expression data, and uses ML approaches to identify known covariates and novel factors.

RT @BaharZirak: Our paper is out!!!.Congrats to all the authors and thank you @genophoria for the opportunity!. Her….

RT @garybader1: Excited to see our strain-specific rat liver single-cell study out now + interesting computational workflow to dissect expe….

Our paper describing the first multi-strain atlas of the healthy rat liver is finally posted @iScience_CP.

RT @cellbrowser: New rat dataset! 🐀 Come explore the rat liver atlas on the UCSC Cell Browser. Special thanks to o….