HUGE shoutout to our amazing team: Wang Wenjing, Liew Wei Lin, Liu Mei Hui, Xueling Sim, and Tai E Shyong!

Colocalization with 12 Asian GWAS traits pinpoints 16 metabolic risk genes, revealing how lifestyle modulates genetic risk at loci like DHRS4L2 and YBX3.

Splicing QTLs (sQTLs) also respond to intervention, with attenuated regulation at T2D risk gene ANK1 and dynamic changes in mitochondrial genes.

We identify hundreds of lifestyle-responsive eQTLs, including context-specific regulation of ACTN3, NEB, and ATP2A1.

Muscle transcriptome rewiring reveals activation of insulin signaling and mitochondrial genes—plus significant shifts in alternative splicing.

We conducted a 16-week lifestyle intervention, which led to ~10% weight loss and improved insulin sensitivity, reshaping body composition and lipid metabolism.

This is the third collaborative paper between the two labs. Looking forward to many projects to come!

And here is the Nature Genetics Cover!

In summary, we established a robust single-cell splicing analysis pipeline and uncovered functional regulatory pathways and novel disease mechanisms.

We also identified and validated a putative causal mechanism.

A genetic variant influencing the TCHP exon 4 inclusion, which putatively affects Graves' disease risk. We validated the variant-to-splicing relationship using a minigene assay.

We identified a pathway from genetic variants -> cis-eQTL (of a splicing factor) -> trans-sQTL (of a key T-cell differentiation marker) -> T-cell proportions (of naive vs. memory T cells). All of these are inferred using a single population-scale single-cell dataset.

Dynamic AS and dynamic sQTLs are widespread along the B-cell development trajectory, some of which may be implicated in disease. For example, PAX5 dynamic isoforms are key determinants of B cell differentiation, and CLEC2D dynamic sQTL is colocalized with autoimmune disease.

We observed significant improvement in exon coverage compared to 3' short-read data (4-6 fold!).

We established a robust pipeline to analyze alternative splicing using 10x 5' single-cell data. Because of "exon painting" effect, ~90% of exons are covered by 5' short-read, despite 5' bias. ~86% of short-read ss are replicated by long-read, likely more with more long-read samples.