search

linnarsson-lab / adult-human-brain

Cytograph version used for adult human-brain analysis
BSD 2-Clause "Simplified" License
71 stars 13 forks source link

Cluster values #3

Open menage95 opened 1 year ago

menage95 commented 1 year ago

Hello Linnarsson lab,

Thank you for such a great resource! I have opened your smaller cluster file in python and have sub-setted some of the clusters I am interested in. I was wondering what the numerical value for each cluster/gene meant? are they a normalised value of some sort?

Thanks! Lucy

slinnarsson commented 1 year ago

They are simple averages of UMI counts, not normalized in any way

rmd13 commented 1 year ago

Hello Linnarsson lab, I am also curious about the normalization in the Genes x clusters: adult_human_20221007.agg.loom. I opened it in hdfviewer and found a TotalUMI table. The data in the main matrix table is float with values distributed from 0 to as high as 7. Many of them are 0.000*. I thought that if a cluster's cell were to be averaged into a number, it is better to normalize the gene level of each cell by the cell's UMI counts, then make an average of all cells inside this clusters. Since UMI counts varies across cells, the cell with lower or higher sequencing depth then will not make the result biased.