# Comparing number of differentially expressed genes (DEGs)

**URL:** <https://discourse.scverse.org/t/comparing-number-of-differentially-expressed-genes-degs/1730>\
**Category:** scanpy\
**Created:** [August 31, 2023, 11:37am UTC](https://discourse.scverse.org/t/comparing-number-of-differentially-expressed-genes-degs/1730 "2023-08-31T11:37:08Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Simon](https://yyz1.discourse-cdn.com/flex035/user_avatar/discourse.scverse.org/simon/32/788_2.png) [@Simon](https://discourse.scverse.org/u/Simon)\
**Post date:** [August 31, 2023, 11:37am UTC](https://discourse.scverse.org/t/comparing-number-of-differentially-expressed-genes-degs/1730/1 "2023-08-31T11:37:08Z")

</div>

Hi,

Is it valid to compare multiple cell types by looking at the number of DEGs with q\<0.05 between two conditions? Can this tell us which cell type shows the strongest effect?

For example, this was done in Allen et al, Cell, 2023, Fig. 4A (10.1016/j.cell.2022.12.010).

Aren’t cluster size or number of unique genes confounders in this analysis? Say cell type A with 5000 cells has much more DEGs than cells type B with only 200 cells. This was briefly touched here:

> [@Understanding differential gene expression analysis](https://discourse.scverse.org/t/understanding-differential-gene-expression-analysis/57/4):
>
> Interesting discussion. Can I ask a question regarding the sampling process. I am fairly new to this tool, and use it now more and like it. Q: If we compare the differential gene expression between one clusters with two conditions - and we have 100 cells from condition\_A and 1000 cells from condition\_B . Condition\_A has a normally distributed expression around 10 and treatment\_B has also a normally distributed expression value around 8 (for the sake of the example, but this brings up another q…

Does iterative random downsampling of the larger cluster help out here? Or are there other suggestions?

Advice is much appreciated!