single-cell accessibility trajectories

chao lu

unread,

Mar 10, 2021, 10:37:00 PM3/10/21

to cicero-users

Hi Hannah!

I am doing single-cell accessibility trajectories based on my scATAC-seq. Currently, I have constructed trajectories with accessibility data. My biggest concern is how to analyze branches in single-cell trajectories, which represents potential cell fate decision points during cell differentiation. I have noticed that monocle3 can help analyze the genes that are regulated around trajectory branch points. Likely, can I extract the peaks that are responsible for the branch? To be more specific, I want to know the differential peaks between branches. I wonder whether there are any solutions to address this problem.

hpl...@gmail.com

unread,

Mar 18, 2021, 2:04:30 PM3/18/21

to cicero-users

Hello,

Yes, I would recommend trying the monocle3 graph_test function, described here https://cole-trapnell-lab.github.io/monocle3/docs/differential/ to find differentially accessible peaks. I have not previously tried this, but don't see why it wouldn't work given that you have sufficient depth for the peaks of interest. Hope this helps!

chao lu

unread,

Mar 18, 2021, 10:21:27 PM3/18/21

to cicero-users

Hi Hannah!

Thanks for your reply! I have another question regarding 'Cicero gene activity scores'. Briefly, as you stated in the tutorial, the input CDS must have a column in the fData table called "gene" which indicates the gene if that peak is a promoter, and NA if the peak is distal. One way to add this column is demonstrated below. However, this might be simple for human or mouse datasets, where the genome structure in my research is a little bit complex. Specifically, some peaks span multiple gene promoter regions, and some peaks simply overlap with multiple gene bodies. In these scenarios, how can I annotate the peak regions ‘cause it might be hard to determine which gene represents the peak. I have enclosed some examples in the attachments

case_1.png

case_2.png

hpl...@gmail.com

unread,

Mar 25, 2021, 3:21:37 PM3/25/21

to cicero-users

Hi,

So in that case, there isn't going to be a computational way to determine which gene is being represented by a promoter peak... I'd recommend naming the promoter peak something that represents that its a combination for the run (gene1_gene2 for example). That should let you generate your results, but I'm afraid you'll need other information to be able to truly assign the score to a gene.