There's been talk about ACE2 and its association with age, sex, smoking etc. as a way to explain #covid19 disease severity. Driving this is intuition that since ACE2 is a receptor for SARS-CoV-2, ⬆️ACE2 => worse disease. My take on this w/@sinabooeshaghi biorxiv.org/content/10.110…
First, if you're going to examine previously published expression data you have to be very careful with ACE2. Expression levels are very low in some organs, e.g. in the lung. This means you have to really worry about how data was processed, normalized, scaled etc.
We looked at mouse single-cell RNA-seq data published by @IliasAngelidis, @lkmklsmn et al. (@fabian_theis and @SchillerLab). Using #scRNAseq from eight 3-month old and seven 24-month old mice we found significantly *less* expression in aged mice. 
This is not a crazy proposition. See, e.g. jamanetwork.com/journals/jamac…
It is also consistent with work in rat from the SARS outbreak days. sciencedirect.com/science/articl…
BUT (capitalized for emphasis) our analysis is of mRNA (not membrane bound protein) and in mouse (not human). We have a specific disclaimer in our preprint.
So please sit down and don't start speculating on how to treat patients. Sadly, there is way too much armchair pharmacology going on.
In human, there is one dataset (@GTExPortal ) that can be examined to look at ACE2 expression variability with age. Several groups have done so, and found small (not significant) decrease in expression with age. However the data is a poor fit for the question.
There are numerous confounders (population, sex, time tissue collected post death etc.) and it's bulk #RNAseq. What see (in mouse) is that not only is ACE2 expressed in only a handful of cell types in the lung, it seems the decrease is specific to ciliated cells.
So it seems that GTEx data is not ideal for answering the particular question of whether ACE2 expression decreases with age in human lung.
Our analysis is full reproducible starting with the reads. This is important! How the data is handled is particularly crucial for this question. See github.com/pachterlab/BP_…
Finally, our preprint was motivated by analysis we performed for a @zoom_us workshop “Bioinformatics for Benched Biologists”. The workshop can be viewed here:
Thanks to several participants for useful feedback.
Thanks to several participants for useful feedback.
