←
Vicky Yao88 posts
Vicky Yao avatar

Vicky Yao

@vyao

assistant prof of CS @RiceUniversity computational biology, disease genomics, data visualization staging.bsky.app/profile/vya…

◷ Joined November 2008▧ 13 Photos and videos
Qiliang Vincent Lai avatar
⟳ Vicky Yao retweeted
Qiliang Vincent Lai@QiliangLai·9 May 2024

Anyone who’s interested come and visit poster 168 so that we can discuss everything there! It’s 7-10 pm tonight at Biondi Hall of #BoG24!

Vicky Yao@vyao9 May 2024

One of the important factors that helped me stop procrastinating writing this thread is that @QiliangLai is at #BoG24 presenting this work later tonight! If you’re around and curious, catch him at poster 186 from 7pm-10pm! n=9

Vicky Yao avatar
Vicky Yao@vyao·9 May 2024

One of the important factors that helped me stop procrastinating writing this thread is that @QiliangLai is at #BoG24 presenting this work later tonight! If you’re around and curious, catch him at poster 186 from 7pm-10pm! n=9

Vicky Yao avatar
Vicky Yao@vyao·9 May 2024

Beyond these neurological case studies, we also apply seismic systematically to a bunch of cell types and other GWAS traits and of course also compare these results to existing methods. 7/n

Post attachment
Vicky Yao avatar
Vicky Yao@vyao·9 May 2024

One of seismic’s most exciting features is that not only can it identify potential associations, but it can also pinpoint the genes driving them! This allows us to glean more insights into which genes (and corresponding processes) help prioritize significant associations. 6/n

Post attachment
Vicky Yao avatar
Vicky Yao@vyao·9 May 2024

Interestingly, if we instead use an AD biomarker GWAS, even though the cohort size is ~20x smaller, we no longer see microglia and instead find many neuronal associations, including ones in the most AD-vulnerable regions, such as neurons from the entorhinal cortex and CA1! 5/n

Post attachment
Vicky Yao avatar
Vicky Yao@vyao·9 May 2024

Looking at Alzheimer’s, we highlight how not only is choice of cell type granularity important, but also GWAS endpoint! This may seem intuitive, but most previous studies focus on the clinical AD GWAS with big cohorts, and like them, we find that microglia is most-associated. 4/n

Post attachment
Vicky Yao avatar
Vicky Yao@vyao·9 May 2024

We develop a novel specificity score that enables seismic to consistently identify disease-associated cell types across different granularities and demonstrate this on GWAS for Parkinson’s disease (where dopaminergic neurons in the substantia nigra are most vulnerable). 3/n

Post attachment
Vicky Yao avatar
Vicky Yao@vyao·9 May 2024

seismic is a scalable, interpretable method to connect GWAS and scRNA-seq. Especially for brain-related studies, previous efforts often use coarse “granularities” for cell types (e.g., cortical neuron) and don't fully take advantage of the more rich delineations of cell type. 2/n

Vicky Yao avatar
Vicky Yao@vyao·9 May 2024

Been behind on tweeting (x’ing?) about recent work, but wanted to highlight a new preprint: seismic, a framework that connects complex traits and diseases to cell types!🌋 This work was a collaboration between my group (@QiliangLai, @dnflsr) and @JP_Roussarie at @BUMedicine. 1/n

Post attachment