search

omnideconv / deconvBench

Comparison of 2nd generation deconvolution methods implemented in omnideconv
2 stars 0 forks source link

mRNA bias in mouse data #12

Closed LorenzoMerotto closed 9 months ago

LorenzoMerotto commented 2 years ago

Hello team! I am currently working on the benchmarking of 2nd gen deconvolution methods on mouse data under the supervision of @FFinotello. In order to build the training dataset I considered a subset of Tabula Muris containing data from Lung, Spleen and Bone Marrow, sequenced with the 10x Genomics protocol.

In this first plot we can see the distribution of the number of distinct genes per cell type: immagine

In this second plot the total number of detected molecules (counts) per cell type is shown: immagine

mlist commented 2 years ago

Looks interesting! What's going on with the erythroblasts? Do they focus so much on hemoglobin or why is there such a large difference from the otherwise relatively consistent pattern between counts and numer of genes...

FFinotello commented 2 years ago

Looks interesting! What's going on with the erythroblasts? Do they focus so much on hemoglobin or why is there such a large difference from the otherwise relatively consistent pattern between counts and numer of genes...

Indeed very nice case, also to test our simulator on. Would be nice to find some spike-in datasets including also erythroblasts. I would say the Census approach would underestimate the total mRNA content in this case