A year ago in Nature Biotechnology, Becht et al. argued that UMAP preserved global structure better than t-SNE. Now George C. Linderman and me wrote a comment saying that their results were entirely due to the different initialization choices: biorxiv.org/content/10.110…. Thread. (1/n)

#mdpientropy IMMIGRATE: A Margin-based Feature Selection Method with Interaction Terms mdpi.com/1099-4300/22/3… #hypothesismargin #featureselection #entropy #IMMIGRATE

6 Free Data Science Books. #BigData #Analytics #DataScience #NLProc #IoT #IIoT #PyTorch #Python #RStats #TensorFlow #Java #JavaScript #ReactJS #GoLang #CloudComputing #Serverless #DataScientist #Linux #Books #Statistics #Programming #Coding #100DaysofCode bit.ly/3cFZbGc

Demographic history impacts stratification in polygenic scores biorxiv.org/content/10.110…

Fun fact: There are about 2,700 trees on the Johns Hopkins University campus, not including forest areas. (📷 Will Kirk / Johns Hopkins University)

"End-to-end neural system identification with neural information flow". Our new PLOS Comp Biol methods paper has been published. doi.org/10.1371/journa… w/ @kateiyas, Luca Ambrogioni, Dr. Yağmur Güçlütürk & Marcel van Gerven. #fmri #systemidentification #visualsystem

Excited to share our latest manuscript on complex trait fine-mapping across diverse populations! This is a collaborative effort across BioBank Japan FinnGen and UK Biobank with mentorship from Hilary Finucane Mark Daly YukinoriOkada. Here is 🧵 1/n medrxiv.org/content/10.110…

Happy to share our new preprint on meta-analysis fine-mapping in the Global Biobank Meta-analysis Initiative Global Biobank Meta-analysis Initiative. Joint work w/ Roy Elzur Wei Zhou under the mentorship of Mark Daly Hilary Finucane. Tl;dr meta-analysis fine-mapping is hard! 1/n medrxiv.org/content/10.110…

How far from a lead SNP should you look for the causal gene? My new paper is out in @BMCBioinfo Thanks for everyone's comments on the previous version, especially those from @tuuli. I reworked that part of the paper and included your prior empirical work pubmed.ncbi.nlm.nih.gov/35527238/

The analysis of single-cell RNA-seq data begins with "normalizing" counts. In a preprint with sina, ingileif & Angel Galvez Merchan, we examine the assumptions and challenges of normalization, benchmark methods, and motivate solutions: biorxiv.org/content/10.110… 🧵 1/

FastIntegration: a fast and high-capacity version of Seurat Integration for large-scale integration of single-cell data biorxiv.org/cgi/content/sh… #biorxiv_bioinfo

New GWAS of T2D (180K cases, 1.2M controls; ~50% non-Eur). Fine-mapping substantially improved with non-Eur samples prioritizing single causal variants at 54% of the 237 loci. Nature Genetics

Using transcriptomic similarity and spatial coordinates, PASTE allows aligning and integrating spatial transcriptomics data generated from adjacent tissue slices. Ron Zeira Ben Raphael nature.com/articles/s4159…

graph-GPA 2.0: A Graphical Model for Multi-disease Analysis of GWAS Results with Integration of Functional Annotation Data arxiv.org/abs/2204.06714

Genetics and genomics play critical roles in health and disease, but what’s the difference between the two? Why are both important for our health? Learn more in our genetics vs. genomics fact sheet! go.usa.gov/xtcWY

A very nice collaboration on multi-ancestry PRS with a combination of penalization and ensemble learning methods. Great collaborations with Jingning Zhang Nilanjan Chatterjee Jin Jin Jianan Zhan Ruzhang Zhao 23andMe @NCIEpiTraining and all others colleagues!

New paper from our lab on methods for bulding high-dimensional predictive models with disparate datasets. Amazing team work by Ruzhang Zhao Prosenjit Kundu and Arkajyoti Saha. arxiv.org/abs/2312.12786 1/5

Thanks for the amazing mentorship of Nilanjan Chatterjee and collaboration with Prosenjit Kundu Arkajyoti Saha !

Our paper on MUSSEL, a method for developing ancestry-specific PRS, is finally out in Cell Genomics : cell.com/cell-genomics/…! Thank you for the > 3-year effort Nilanjan Chatterjee Jianan Zhan Haoyu Zhang Ruzhang Zhao Updated software: github.com/Jin93/MUSSEL Feel free to try it out!

Our paper on PROSPER, a new multi-ancestry PRS method with penalized regression is out on Nature Communications ! Check here nature.com/articles/s4146…. Thanks to everyone offering help along the way!! Nilanjan Chatterjee Haoyu Zhang Jin Jin Jianan Zhan Ruzhang Zhao Cheng Ma and PRIMED Consortium