How does tumour heterogeneity arise? How can we predict cancer cell plasticity? In 2 studies, we trace #glioblastoma heterogeneity to a spatial cancer cell trajectory w. atlassing & predict plasticity w. snRNA/ATAC+deep learning 🧵

RT @dilekopter: Happy to share our new study from my PhD on the spread of farming across Anatolia and into the Aegean and then into Europe.….

RT @hana_aliee: I'm honoured to join @CRUK_CI and @Cambridge_Uni as a Junior Group Leader of AI for Cancer. My group will focus on reasonin….

RT @DervisSalih1: We are happy to be seeking three post-docs across Edinburgh Napier, Glasgow & UCL in an exciting project to investigate m….

RT @sangerinstitute: Huge congratulations to Sam Behjati who has been elected as a Fellow of the Academy of Medical Sciences 🎉. Sam’s work….

RT @bayraktar_lab: How does tumour heterogeneity arise? How can we predict cancer cell plasticity? In 2 studies, we trace #glioblastoma het….

RT @ahmet_rifaioglu: Great effort by @bayraktar_lab, @StatGenomics & @MoritzMall labs in unravelling GBM cell state variability and spatial….

RT @MoritzMall: What keeps cancer cells in check—and can we tame hem? In 2 collaborative preprints, we trace glioblastoma cell transitions….

RT @dana_peer: Truly excellent and important work.

Grateful to >60 authors for their amazing work + @WellcomeLeap (special thx to @jrswedlow for leading DeltaT!) & @cziscience for their support. @emblebi @cricklab @Cambridge_Uni @DKFZ @sangerinstitute.

Finally, you can access & navigate almost all of our GB single cell and spatial data in our portal @ built with webatlas2 (.

There are many implications & LOTS of detail in the preprints (e.g. mesenchymal cancer states is really a gliosis/hypoxia response & GB tumour subtypes are likely confounded)

Taken  together, these 2 studies start to unveil what I call the “rules” of glioblastoma. Despite their incredible complexity, these tumours are governed by specific cellular & tissue mechanisms. We believe/hope this understanding will be a foundation for new therapies

In GB, as opposed to unrestrained plasticity of cancer cells, we predict transition highways and barriers aligned with paper 1. And we mechanistically validate a transcriptional repressor safeguarding neuronal like cancer states!

scDORI enables us to examine GRNs in a continuous manner across cellular trajectories. Hence, you don’t just map GRNs / TF regulators, you can actually infer cell plasticity and predict the future state of a given cell

Here we developed a new deep learning based GRN inference framework: scDORI. See the thread from @manusaraswat10 for more details but few highlights here

In paper 2, we leverage our massive snRNA-ATACseq data to predict cancer cell state transitions & trajectories from gene regulatory networks. Led in @OliverStegle @MoritzMall labs by @manusaraswat10 Laura, Elisa + Tannia, @FaniMemi

This picture is very consistent with a concurrent study from @slinnarsson lab (not active on X) - check their paper out:

Hence, we define a stereotyped trajectory of cancer cells at the heart of GB heterogeneity - unifying previously defined cell states & tumour subtypes

Finally, cancer cells are not riding alone. We found that this cancer cell trajectory is intimately linked to myeloid heterogeneity and unfolds across regionalised myeloid signalling environments.

This led to a wonderful side quest: we mapped tumour subclones in space! We developed SpaceTree to deconvolve tumour subclones in Visium data and found that they are finely spatially intermixed (within Visium spots) in the tissue!