Excited to announce the Cook Lab will be opening at the Ottawa Hospital Research Institute Jan 2024. We'll be focusing on intratumoural heterogeneity and therapy resistance in ovarian cancer
Eager to work more with the ovarian cancer research community!
Thrilled to see that our paper using MULTI-seq to compare transcriptional dynamics of the EMT across contexts is now up on
@NatureComms
. scRNA-seq of 12 EMT time courses (4 lines, 3 inducers) and screened effects of 22 kinase inhibitors on each combination
Excited to share some new work w/
@bvanderhyden
. Used archetypal analysis to learn features of epithelial-mesenchymal plasticity from scRNA-seq data of 160 tumours spanning 8 cancer types. Transcriptional census of epithelial-mesenchymal plasticitiy:
See this often enough that I think it warrants a tweet. Be aware that if you feed Seurat's FeaturePlot >2 custom colors, the values in the plot's scale no longer reflect normalized counts. Can add ggplot2 code to get around this is you still want to use FeaturePlot
Excited to share this work! We wanted to compare the transcriptional dynamics of the EMT across contexts, so we used MULTI-seq to generate scRNA-seq data for 12 EMT time courses and 16 kinase inhibitor screens. With replicates, 960 samples in total.
Excited to share our comparison of Xenium and CosMx data! I feel like a lot of labs/core facilities have made similar observations internally, so we're really glad to put this out for the broader community. 1/n
I’m going to quietly leave this here…but not without a shoutout to
@DavidPCook
,
@kirkbjensen
and
@Kellieiswise
for their AMAZING work. 🙌
A Comparative Analysis of Imaging-Based Spatial Transcriptomics Platforms
Great that we're starting to see some data comparing these platforms head-to-head. And kudos to those making the data public. I think we do a disservice trying to boil the comparison down to a couple summary metrics. Eg. dove into this breast cancer data for an hour 1/n
@LGMartelotto
, further to our conversation comparing $TXG 10X Genomics Xenium and $NSTG NanoString CosMx Spatial Molecular Imager for spatial transcriptomics (ST), we have new data to consider from the lab of Dr Yutaka Suzuki from the University of Tokyo.
Dr Suzuki has long been
Thrilled to announce that I've received a
#BantingCanada
fellowship from
@CIHR_IRSC
. Excited to be working on cell plasticity with Jeff Wrana at the Lunenfeld-Tanenbaum Research Institute
@SinaiHealth
.
Successful thesis defense out of the way!
@bvanderhyden
, thanks for such an incredible PhD experience. And thanks to
@ItaiYanai
and my other evaluators for taking the time to read my work and give thoughtful feedback. Shame we couldn’t celebrate in person!
Having fun with the Tabula Muris
@10xgenomics
datasets in scanpy. 55k cells runs smoothly on laptop. Force-directed graph of 12 tissues looks like some abstract art
Thank you 🙏. Definitely an honour to receive the Worton Researcher in Training Award. Couldn’t have done any of it without the supportive environment at
@OttawaHospital
/
@uOttawaMed
or my fantastic colleagues and mentors (a particularly special thanks to
@bvanderhyden
)
Congratulations to
@DavidPCook
on receiving the Worton Researcher in Training Award in recognition of his outstanding
#cancer
research achievements and pioneering new techniques!
Learn how this rock star researcher uses big data to answer big questions:
Great perspective from the
@fabian_theis
group on machine learning for perturbational single-cell genomics. Parts about learning a perturbation latent space remind me of Cleary/Regev's points about value of designing random experiments for inference
Excited to talk about this. I came to this project just trying to inform myself to help my institutes with purchase decisions. I have no inherent bias for either platform (and they certainly don’t pay me 🙃). Just want to provide transparent, practical insight into the data
#Singlecellspatial
imaging technologies empower new insights—but how do the different platforms stack up? Find out in a webinar where leading spatial researchers compared Xenium In Situ and NanoString CosMx and see which came out on top in their hands:
Multimodal Analysis of Composition and Spatial Architecture of Human Squamous Cell Carcinoma. scRNA-seq + spatial transcriptomics + in vivo CRISPR screening of SCC xenographs. Very cool work.
@arjunrajlab
I find lazy critique like this is getting reinforced in grad classes. Told to evaluate a paper and get marked for participating in discussion. Everyone scrambles to find anything to say (stats, replicates, protein vs rna) and come to believe that this is how to critique papers
Single-cell folks! Any suggestions for approaching multiome sample prep for large, staggered batches of organoids? ~20 samples per timepoint. Recommendations for fix(?), freeze, and parallel processing (v-bottom plate hacks or something?) appreciated! CC
@LGMartelotto
@Sc_Ninjas
Head-to-head alignment of some new data with new cellranger update. Comparing v6 and v7 default, as well as the optimized mm10 reference from
@Allan_HPool
. Median gene ~600 higher. Median UMI ~3k higher. Curious about gene-level diffs with optimized ref.
If anyone is using the Seurat + Monocle wrapper vignette as a reference for their own trajectory inference, be cautious how the assays are being transferred. In the vignette, after seurat > CDS > seurat, the original integrated assay gets put into "RNA"
@achamess
@LGMartelotto
@tangming2005
Usually have key questions at the onset to point, but usually goes 1) freestyle exploration in different directions that seem interesting, 2) scripts become wildly disorganized, 3) burn it down, re-write everything in some logical order based on interesting angles from (1)
Gene expression changes in response to a stimulus are often interpreted as a change in "cell state". Wonder how often exp changes are to *maintain* cell function under different environmental pressures. Don't know of any examples, but interesting thought. Still a "state" change?
Hexbinning gene expression on scRNA-seq embeddings is a nice way to smooth out noisy/sparse signal for visualizations, particularly when the embeddings are heavily overplotted.
Beautiful work. Really want to work with joint data like this. Chromatin state change precedes gene expression. Incongruence of profiles can be used to predict cells' future state. Congrats
@maple1989
et al!
Does chromatin and gene expression define the same cell states? To answer this, we developed a new technology, SHARE-seq to bring measurements of regulatory mechanisms to single cells.
For visualizing subtle shifts in an embedding. Toying around with visualizing diff in density estimates between two conditions. Kind of an exaggerated example, but could be a cool way to see diffs in one plot instead of 2-color dot confetti. Also controls for diff cell counts
Had a lot of fun working on this with
@IvanaMizikova
and
@HurskainenMaria
from
@LabThebaud
! Used MULTI-seq for scRNA-seq on 36 mouse lung samples throughout normal and impaired late lung development. BPD is associated with progressive phenotypic changes in all cell types
Excited to see our new opinion article in
@trendscancer
! Having struggled to establish unifying molecular principles of EMP in cancer, I've been exploring new conceptual frameworks to help understand its emergence and consequences.
Hey single-cell folks. Anyone ever have two similarly sized groups of cells that differ in total UMI counts by large amount (avg 4k vs. 25k) and DGE is largely translation genes higher in group with more UMIs without much else? Weird technical thing? ATAC data doesn't show split
@ItaiYanai
I really feel for all the mid-stage grad students that are doing amazing work but start to doubt themselves because they’re losing award competitions to students that maybe fell into a near-complete project and published it. The CV lag is hard for the first 5 years or so.
⭐️⭐️⭐️We are pleased to announce the winners of the 2022 OICR Rising Stars Awards⭐️⭐️⭐️
Congratulations to all six awardees this year! Check out more info about them below!
Thank you to everyone who applied and we hope to see your application for our next round of awards! (1/6)
Why do my own work when I can spend my day visualizing others’. Force-directed graph of >20k single cells from the recent E8.25 murine embryo paper from the Marioni group. Paper:
Have had a chance to try out bigSCale2 over the last couple months. I've largely been using it for calculating and comparing TF centrality across the regulatory networks it constructs. It's been performing great for me!
RELEASED! Our updated
#SingleCell
Analysis Framework
#bigSCale2
().
1. Now in
#R
2. Including new feature: Super-clustering
3. New module: Gene Regulatory Network inferrence.
4. Ready-to-use Index Cell (iCell) datasets:
Predicting a cell's future state w/ RNA velocity. Excited to see this finally published! Congrats
@GioeleLaManno
! Time for me to learn enough Python to actually use this efficiently 😅.
Check it out this Wednesday from 2-3pm EST! Want to emphasize that this isn't just platform promo/attack. For the sake of our own research, we've been genuinely trying to understand how these platforms perform, where they excel, and where they fall behind.
#Singlecellspatial
imaging technologies empower new insights—but how do the different platforms stack up? Find out in a webinar where leading spatial researchers compared Xenium In Situ and NanoString CosMx and see which came out on top in their hands:
Am I the only one that is visually drawn to dot size more than colour in the default dot plots like this? I find them awfully hard to interpret quickly when the two variables are not correlated.
Personal site launched at ! Just a fun thing to toy around with. Hope share analyses and thoughts through blog posts and slide decks from talks. First post up: Scanpy, The Mouse Atlas, and Baby Steps Into Python.
@NinaSteele17
@Momademia
Our girl went through a phase of having a “witching hour” for a couple hours each night. Perseverance, tagging partner in, along with swaddling, rocking, shushing was our approach. Also had GI issues if put down too soon after feeding. Had to keep her upright for 30mins or so
The Stem Cell Network and
@OIRMnews
are pleased to offer the RNA-Seq Analysis workshop once again! Workshop Dates: October 16, 17 & 18, 2019. Application deadline: August 28, 2019. For application & travel award details click here:
Paper for side project recently submitted to 6th journal, and received 11th scholarship rejection over 4 years of grad school today (though I've been lucky to have received an award 3/4 of those years). Learning that coping with rejection is important in science (and hard to do!)
Interesting way to integrate single-cell data from multiple experiments. Seems to perform well - Panoramic stitching of heterogeneous single-cell transcriptomic data
If you’re interested in multiplexing scRNA-seq samples (which you should be!), I strongly recommend taking advantage of this. Labeling steps are simple and fast. Happy to share my experience if you have questions!
I already forget the paper from a few months ago that first did density plots like this, but I'm loving using them to look at treatment responses in scRNA-seq data
Fun project
@lrncarter
@bvanderhyden
. “Stemness” of simple epithelial cells is discussed a lot, but perturbations that increase it do not promote defined responses and stemness markers are inconsistent. Suspect stemness is a fluffier concept than we give it credit for
@bvanderhyden
@DavidPCook
@lrncarter
et. al identify heterogeneity in the transcriptional profiles of ovarian surface epithelial stem cells induced under different conditions, suggesting transient induction dependent on environmental cues.
@Sanbomics
@scverse_team
Definitely useful. Appreciate Seurat’s shuffle=T for convenience. Feel like reproducible shuffling would be a sensible default actually
Fig 1c: PCA coloured by treatment
Fig 2a: Same PCA coloured by cluster
Reviewer: "Although you used clustered the data, they look very much like the PCA in Fig 1".
Me: ...
@uOttawaBMI
@lisadambrosio2
Mireille has been a big part of the uOttawa research community since I started as a summer student over 4 years ago. Great to see her join the faculty. Looking forward to seeing her program develop!
@rmassonix
@Sc_Ninjas
Oh interesting! Didn’t know this existed. Wonder how it works under the hood. I always subsetted the data and reprocessed because I figured there was value in defining cluster-specific variable genes, re-running PCA on those, and using that for clustering.
MULTI-seq - Lipid anchored indices for fast, non-perturbative multiplexing of scRNA-seq samples. I was working on an identical approach, so a little bummed about that, but happy to see the method work so well!
@melodycoutman
1) Take initiative in steering your own project. I find many students only do what their supervisor wants and quickly become detached from the project and stop enjoying the experience. 2) Network/make friends with other trainees outside your lab. Great for venting & collaboration
Just had a chance to try out Harmony for aligning multiple single-cell datasets. Works remarkably well and is super fast! . >30k cells and ran on laptop in <5mins. Before and After:
At the airport drinking my last cup of Tim Hortons for a week. Excited to head off to the
@NatureConf
“The Tumour Cell: Plasticity, Progression, and Therapy” in NYC
#Tumour19
Graph layout of embryonic organ explants from genetic mouse model. Nice reminder that effects of gene/pathway manipulation can be quite variable across a single organ: depletion of one cell type, divergence of two others, and no noticeable effect across the rest
@rmassonix
I appreciate and agree with all those points, but I suspect that, for better or worse, many (>50%?) people analyzing and publishing sc data have a casual background in coding at best. As it is now, the python ecosystem is simply too impenetrable to these users.
It’s finally out! How to use scRNAseq to dissect schizophrenia. Thanks to our great collaborators in the Sullivan lab and others.
@n_skene
@slinnarsson
@PGCgenetics
Been thinking about this one for the last two days. Pretty messed up that the field can sing praise about this work and it gets outright rejected. Can’t believe how happy I am the preprint exists. Running Velocyto as I type
Reviews back for our RNA velocity paper: only two reviews (?!). One referee calls it ”a transformative contribution to the field”, the other ”very exciting” (but nitpicks about splicing kinetics). Paper rejected outright 😒. WTF.
Also, quantifying transcripts per X isn't very informative when the probesets are wildly different. The CosMx panel includes some highly expressed genes that are avoided in the 10x probes. In the CosMx data, MALAT1, XBP1, and TPT1 account for 25% of all transcripts
Spending more time in Python recently, I've really come to appreciate the documentation standards for R packages. sci-kit learn aside, I miss having thorough vignettes
@strnr
I’ve been wondering lately about how many of the distinct populations/“cell types” inferred from tSNE are either continuous phenotypes to other populations that get detached by the algorithm or even just artificial breaks in the underlying structure