Skip to contents

This page collects material outside the package vignette tutorials: the working methods manuscript, seminar slides, and canonical URLs. The in-package articles under Vignettes document implementation, simulation, and mathematical detail; start here when you need the paper PDF or talk deck.

Resource Link
Source repository bastienchassagnol/DeCovarT
Methods preprint arXiv:2309.09557 (Chassagnol et al. 2023)
Package site bastienchassagnol.github.io/DeCovarT
Function call network decovart_function_network.html

Methods manuscript

The manuscript describes covariance-aware deconvolution of bulk transcriptomes as a multivariate Gaussian convolution. Cell-type proportions lie on the simplex; they are recovered by constrained maximum likelihood with analytic gradients and Hessians. The PDF below is the working build of article/main.pdf (Springer Nature sn-jnl template).

Open full screen or download a copy.

Doctorants seminar slides

This embeds the Quarto reveal.js deck from the doctorants deconvolution seminar (sources under slides/ in the repository). Pattern inspired by GTBioSS_slides.qmd and the quarto-webr slide-embed example.

Open the deck full screen.

References

Chassagnol, Bastien, Grégory Nuel, and Etienne Becht. 2023. DeCovarT, a Multidimensional Probabilistic Model for the Deconvolution of Heterogeneous Transcriptomic Samples. arXiv. https://doi.org/10.48550/arxiv.2309.09557.