Doubts about mouse RNA-Seq single cell data

Science
, , , , ,
1

Hi Allen Community,

I’m trying to compare mouse RNA-Seq single cell data [WHOLE CORTEX & HIPPOCAMPUS - 10X GENOMICS (2020) WITH 10X-SMART-SEQ TAXONOMY (2020)] with my own proteomic data. My goal, among other things, is to see the expression level of genes that we consider part of the PSD in the different hippocampus cell types.

I’ve come across two issues about the data, which you offer in the form of trimmed means [Gene Expression by Cluster, trimmed means file] and medians [Gene Expression by Cluster, median file]. The first problem is the units. In your data, for example, type 273_CA1-ProS for the Xkr4 gene is expressed in 7.273118 on trimmed mean and 731 on median. Why are the units different? What is trimmed mean and median for you?

The second problem is the difference between the number of expressed genes according to trimmed means or median data. Following the above example, 273_CA1-ProS expresses 7517 genes according to trimmed mean data and 4528 according to median data. As I understand it, for the trimmed means disregard the lower and upper limits of the data, by 25% respectively, so it would be logical that there were more genes in the median data. For what reason is it the opposite, and is there more genes in the truncated mean data?

I have reviewed the readme file you offer and I can’t find the solution, not even in this forum. I also reviewed the help documents you provided and couldn’t find the answer. I hope you can help me.

Thank you so much,

Albert.

2

Hi, Albert

I have downloaded the file for medians, and Xkr4 gene expression is 7.31 for median. So the difference between median and trimmed mean is reasonable. As regarding to the number of detected genes at trimmed means and medians, as genes expressed in >=25% cells are detected for trimmed means, and genes expressed in >=50% cells are detected for medians, there are always more genes detected for trimmed means than medians. Due to extensive drop off in 10X, such differences can be fairly big.
Hope that I have addressed your question, and feel free to reach out for more questions!

Thanks for your interest in our dataset.

3

Hello zizheny,
first of all sorry for taking so long to respond. I appreciate your reply.
About the units, it’s my fault. I didn’t notice that my Excel doesn’t interpret points as delimiting decimals.
On second question, from what you tell me I don’t understand exactly how you calculate the trimmed mean and the median. Could you explain it in detail?
Thank you so much,
Albert.

4

Can someone help me with this, please?

5

Hi, Albert

Trimmed means are computed by removing the top 25% and bottom 25% cells, and compute the mean for the remaining cells.