←
Ross Hardison79 posts
Ross Hardison avatar

Ross Hardison

@rosshardison

Biochemist, genome scientist, teacher at Penn State Univ. Genomic approaches to understand regulation of gene expression. Enjoy guitar and hiking.

◷ Joined February 2017▧ 20 Photos and videos
Ross Hardison avatar
Ross Hardison@rosshardison·Sep 15

My colleagues in genomics and statistics just published the method RETROFIT for reference-free deconvolution of spatial transcriptomics. The editors at Nature Communications just highlighted it as one of the best papers! nature.com/collections/cfhfe…

Ross Hardison avatar
Ross Hardison@rosshardison·18 Jul 2024

Our sets of candidate cis-regulatory elements (cCREs) in human and mouse blood cells are powerful resources for further research. Their high enrichment in genetic variants associated with blood cell traits suggests they play roles in developing those traits.

Post attachment
Ross Hardison avatar
Ross Hardison@rosshardison·18 Jul 2024

Focusing on NMF factor 3, associated with positive gene regulation in erythroid cells, brings out the distal element as well as the well-known proximal regulatory elements in the mouse GATA1 gene.

Post attachment
Ross Hardison avatar
Ross Hardison@rosshardison·18 Jul 2024

Continuing: The correlation matrix is complicated, with states associated with positive or negative regulation giving high correlations (red in the matrix). Decomposition by nonnegative matrix factorization (NMF) separates these different patterns into interpretable NMF factors.

Post attachment
Ross Hardison avatar
Ross Hardison@rosshardison·17 Jul 2024

Our sets of candidate cis-regulatory elements (cCREs) in human and mouse blood cells are powerful resources for further research. Their high enrichment in genetic variants associated with blood cell traits suggests they play roles in developing those traits.

Post attachment
Ross Hardison avatar
Ross Hardison@rosshardison·17 Jul 2024

Focusing on NMF factor 3, associated with positive gene regulation in erythroid cells, brings out the distal element as well as the well-known proximal regulatory elements in the mouse GATA1 gene.

Post attachment
Ross Hardison avatar
Ross Hardison@rosshardison·17 Jul 2024

The jointly learned model provided a common “alphabet” of epigenetic states that allowed us to construct the interspecies correlation matrix. We can compare homologous loci between species using this state “alphabet” without the limitations of sequence conservation.

Post attachment
Ross Hardison avatar
Ross Hardison@rosshardison·17 Jul 2024

The process starts by jointly learning epigenetic states from data in many blood cell types in human and mouse, adopting the Bayesian approach in IDEAS to learn the model jointly. Each 200bp bin in mouse and human was assigned to a state in each cell type.

Post attachment
Ross Hardison avatar
Ross Hardison@rosshardison·17 Jul 2024

Interspecies DNA sequence alignments can find a subset of gene regulatory elements, but they miss about 40% of them in mammals. This work from @Guanjue et al compares patterns of epigenetic states across cell types between species to find some missed elements.

Post attachment
Ross Hardison avatar
Ross Hardison@rosshardison·8 Apr 2022

Love that medley of “Jack Straw”, “China Cat Sunflower”, and “I Know You Rider”.

That album is just remarkable, and while it was modified in the studio, that it was essentially a live album is amazing. ‘He’s gone’ from that album is just perfect.