Lightnews — Scholar-powered news

Lucie GB. @lgaspardboulinc.bsky.social · 22d

Thanks @mbarse.bsky.social for showcasing our review in your Journal club. I hope the review and the additional ressources will be helpful in your work or team!

Manisha Barse @mbarse.bsky.social · Sep 2

Hi everyone! Check my journal club talk on the latest review paper: Cell-type deconvolution methods for spatial transcriptomics 🧬

🎥 www.youtube.com/watch?v=xDjF...
🛝 speakerdeck.com/manishabarse

@lieberinstitute.bsky.social #Bioinformatics #SpatialTranscriptomics #Cell-TypeDeconvolution

[2025-08-13] Journal club: Cell-type deconvolution methods for spatial transcriptomics

YouTube video by Leonardo Collado Torres

www.youtube.com

1

Lucie GB. @lgaspardboulinc.bsky.social · Jul 7

Can't wait to be at @jobim2025.bsky.social #JOBIM25 to present a short demo on the companion ShinyApp (shorturl.at/Os7kB) of our review "Cell-type deconvolution for spatial transcriptomics" recently published in @natrevgenet.nature.com ! #spatial #genomics

Deconvolution toolbox for spatial transcriptomics

shorturl.at

2 4

Lucie GB. @lgaspardboulinc.bsky.social · May 19

🥰Congratulations to all authors and particularly to my co-author Luca_Gortana and Florence Cavalli who supervised our work with Thomas Walter and Emmanuel Barillot. And also thanks to the reviewers whose comments improved the review.
@institutcurie.bsky.social @inserm.fr @psl-univ.bsky.social

Lucie GB. @lgaspardboulinc.bsky.social · May 19

⏭️Looking ahead, numerous opportunities exist to improve cell-type deconvolution. We propose 5 directions of development. New methods will certainly emerge and we hope to maintain our web-tool with the most recent works you can submit here: forms.gle/fuUDhgMrfYvY...

1

Lucie GB. @lgaspardboulinc.bsky.social · May 19

🧐We finally give some clue on #validation of deconvolution results which remains tricky for both developpers and users. We also mention how deconvolution results can be integrated in ST data analysis pipelines to unravel cell-type spatial distribution in complex tissues.

1

Lucie GB. @lgaspardboulinc.bsky.social · May 19

😬From that point, picking a method that best fit a project seems impossible. We thus provide key questions to ask to select methods(s) that may work and a dynamic web tool to browse through all these methods. Check it out! cavallilab-curie.shinyapps.io/Review-Deconvo

cavallilab-curie.shinyapps.io

1

Lucie GB. @lgaspardboulinc.bsky.social · May 19

🚀Going deeper, we demonstrate that #deconvolution is now a multi-modal data integration technique with most frameworks using #singlecell data, #spatial coordinates or tissue image. We delineate the diversity of approaches to integrate these data types.

1 1

Lucie GB. @lgaspardboulinc.bsky.social · May 19

📚We particularly summarize their characteristics regarding input, output, programming languages (#R and #Python mostly), frameworks and presence in independent #benchmarkings. It shows a very high diversity in all these parameters and the need for new benchmarks.

1

Lucie GB. @lgaspardboulinc.bsky.social · May 19

We then take a deep-dive into the mathematical #frameworks on which deconvolution methods rely. We review more than 60 methods and propose a classification depending on their formulation of the deconvolution problem.

1

Lucie GB. @lgaspardboulinc.bsky.social · May 19

We first present the overall workflow of cell-type #deconvolution for #spatialtranscriptomics from input to output. We highlight a first dichotomy between methods depending on wether they use #singlecell data and also pinpoint the novel use of tissue image as a new modality.

1

Lucie GB. @lgaspardboulinc.bsky.social · May 19

🚨Thrilled to finally see our review "Cell-type deconvolution methods for spatial transcriptomics" published in @NatureRevGenet
@SpringerNature

nature.com/articles/s41... (read free: rdcu.be/el0Ka).

Check out our web-tool: cavallilab-curie.shinyapps.io/Review-Deconvo