Excited to share our new preprint! We map the spatiotemporal dynamics of therapy response & minimal residual disease in BRCA1;p53-deficient breast #cancer using #MultiOmics. We identify drug-tolerant EMT cancer cells, offering new therapeutic targets. 🔗 https://t.co/VostKuvNOw

Take a look inside a SARS-CoV-2-infected mouse lung with X-Pression! Our 3D reconstructions reveal changes in gene expression programs across the entire organ, leveraging ST and micro-CT. Check out the details in our preprint: https://t.co/1QZS4kVaKW #SpatialTranscriptomics

Excited to share our new #biorxiv preprint! X-Pression combines deep learning with micro-CT to reconstruct 3D #spatialtranscriptomics from a single 2D section. A powerful, cost-effective way to explore tissues in 3D! Link: https://t.co/1QZS4kVaKW #SARSCoV2-infected lung below!

Check out our GitHub to try Chrysalis! https://t.co/YErV4zTXMj 🧵 (7/7)

Chrysalis can be easily integrated into any workflow based on #Scanpy and works seamlessly with multi-sample datasets. A huge thanks again to Jelica Vasiljevic, Kerstin Hahn, @RottenbergSven, and @alvaldeolivas for their invaluable contributions. 🧵 (6/7)

Leveraging deep learning, we integrated morphological features from H&E images to uncover tissue compartments defined by morphology and expression. We also tested Chrysalis on multiple platforms, such as #VisiumHD, #SlideSeq, and #StereoSeq. 🧵 (5/7)

We benchmarked Chrysalis against other matrix decomposition and spatial domain inference methods with a focus on the #Visium platform, using both in silico and real-world datasets, demonstrating superior performance. 🧵 (4/7)

By fitting a simplex to the low-dimensional expression space, tissue compartments emerge as functionally distinct cellular niches. 🧵 (3/7)

Still using NMF? Try Chrysalis! Chrysalis combines spatially variable gene detection with archetypal analysis to decompose the gene expression matrix. The tissue compartments can be visualised simultaneously, with a unique twist. 🧵 (2/7)

Super happy to announce that Chrysalis is finally out! Our new reference-free method identifies cellular niches in #spatialtranscriptomics with robust performance. Paper: https://t.co/XXPwC1Ze6D 🧵 (1/7)

Chrysalis, a machine learning-based framework, accurately infers cellular niches and underlying gene expression programs in the tissue from spatial transcriptomics data https://t.co/B9TJN0Hlb6 @Rockdeme @RottenbergSven @alvaldeolivas

Happy to share our latest finding on H2AX promoting replication fork degradation and chemosensitivity in BRCA-deficient tumors https://t.co/q6Ydry042k

Excited to share that our paper “Profiling the heterogeneity of colorectal cancer consensus molecular subtypes using spatial transcriptomics” is now published in NPJ Precision Oncology! #CancerResearch #SpatialTranscriptomics #PrecisionOncology @Nature_NPJ https://t.co/QgrLLzYR0D

Check out our Github! https://t.co/YErV4zTXMj 🧵 (6/6)

Chrysalis can be easily integrated into any workflow based on #scanpy and works on multi-sample datasets. Huge thanks to Jelica Vasiljevic, Kerstin Hahn, @RottenbergSven, and @alvaldeolivas for their invaluable contributions. 🧵 (5/6)

Leveraging deep learning, we integrate morphological features from the H&E image to discover unique tissue compartments. We also demonstrate the versatility of chrysalis using #Visium, #Slideseq, and #Stereoseq data. 🧵 (4/6)

We compare chrysalis with other matrix decomposition/spatial domain inference methods, like NSF and MEFISTO, relying on #cell2location and #Xenium reference data. 🧵 (3/6)

Chrysalis combines spatially variable gene detection with archetypal analysis to decompose the gene expression matrix, and employs MIP for simultaneous compartment visualization. 🧵 (2/6)

Super excited to share the preprint of chrysalis, our method that identifies tissue compartments in #spatialtranscriptomics without external reference. Preprint: https://t.co/WlJzMcj92k 🧵 (1/6)

Interested in a PhD project on radioresistance in Bern? Check this out: