←
Shreejoy Tripathy5,550 posts
Shreejoy Tripathy avatar

Shreejoy Tripathy

@neuronJoy

Associate Prof doing bioinformatics and computational neuro at @UofT and @CAMHresearch. Tweets about neurons, ephys, and single-cell RNAseq.

◷ Joined February 2009▧ 167 Photos and videos
Shreejoy Tripathy avatar
Shreejoy Tripathy@neuronJoy·Nov 5

I feel like brisc is best suited for situations where you have: 1) A LOT of data (say, like 2M+ cells), 2) have a huge, dense spatial omics dataset (like Atera?), 3) are doing a large snRNAseq meta-analysis (like what we did here: biorxiv.org/content/10.64898…).

Shreejoy Tripathy avatar
Shreejoy Tripathy@neuronJoy·Nov 5

brisc is transformative because it makes traditionally "slow" parts of the single-cell pipeline, like loading data and label transfer, super fast (like minutes on massive datasets). so it's trivial to test out multiple references for label transfer, etc.

Shreejoy Tripathy avatar
Shreejoy Tripathy@neuronJoy·Nov 5

my lab has been lucky to be using brisc over the last few months as early beta testers and I can say that it's truly transformative for you can do with single-cell data.

Thrilled to release the biggest project of my career: brisc, a full-fledged single-cell toolkit that runs multithreaded at 100x the speed of Scanpy and Seurat on tens of millions of cells. brisc.run

Shreejoy Tripathy avatar
Shreejoy Tripathy@neuronJoy·Nov 2

My lab does something like a mix of traditional lab meetings (1 presenter for one hour) and continual update lab meetings (everyone briefly presents, 3-5 minutes per person). we do these in separate weekly meetings and it works fairly well.

One of the most unique things to each lab is how they run their lab meetings. Ours is a bit different from most labs. Everyone is up every week. Yes everyone , not once in 2 months update but rather weekly updates by all. Let me break it down why this works for us 🧵

Post attachment
Shreejoy Tripathy avatar
Shreejoy Tripathy@neuronJoy·Sep 29

who knew that countless hours playing starcraft in highschool would pay off handsomely in this era of agentic ai. still need more pylons though.

Panos Roussos avatar
⟳ Shreejoy Tripathy retweeted
Panos Roussos@panos_roussos·Sep 26

Thrilled to see PsychAD published! 🧠 A journey begun in 2019: 60+ investigators, 8 institutions, 1,494 human brains—illuminating brain aging and disease. Supported by NIA. Made possible by the generosity of donors and families. psych-ad.org/ nature.com/collections/jfibi…

Shreejoy Tripathy avatar
Shreejoy Tripathy@neuronJoy·Sep 22

We release full DE + abundance results for every cortical cell type, all code, and harmonized cell-type annotations for every cell in the 8 public datasets github.com/stripathy/scz_cel….

Shreejoy Tripathy avatar
Shreejoy Tripathy@neuronJoy·Sep 22

Psychiatric disorders have long been assumed to involve neurons that malfunction rather than disappear. Our study challenges that view and nominates an intrinsically vulnerable population as a target for neuroprotective therapies.

Shreejoy Tripathy avatar
Shreejoy Tripathy@neuronJoy·Sep 22

This matters because SCZ genetic risk (incl. HCN1) concentrates in the most depleted Sst subtypes. They include primate-specialized types with no clear rodent counterpart, like Cajal's double-bouquet cell. And the same subtypes are among the first neurons lost in Alzheimer's.

Shreejoy Tripathy avatar
Shreejoy Tripathy@neuronJoy·Sep 22

We found that both camps were right. Sst interneurons as a whole show broad transcriptional change in SCZ, including less SST mRNA. But a specific subset of upper-layer Sst subtypes is also reduced in number, and that depletion replicates in intact tissue.

Post attachment
Shreejoy Tripathy avatar
Shreejoy Tripathy@neuronJoy·Sep 22

We harmonized 7 snRNA-seq datasets of post-mortem prefrontal cortex onto one fine-grained taxonomy, meta-analyzed expression + abundance, then replicated in an 8th cohort with Xenium spatial data.

Post attachment
Shreejoy Tripathy avatar
Shreejoy Tripathy@neuronJoy·Sep 22

For 20+ years, the field has argued over whether cortical inhibitory interneurons are lost in schizophrenia, or persist in an altered molecular state. Single-cell data could settle this, but each study alone has been too small.

Post attachment
Shreejoy Tripathy avatar
Shreejoy Tripathy@neuronJoy·Sep 22

New preprint from our lab: "Selective depletion of upper-layer somatostatin interneuron subtypes in schizophrenia." The largest case/control single-cell analysis of the schizophrenia brain to date. 🧵 biorxiv.org/content/10.64898…