Entering edit mode
18 months ago
Assa Yeroslaviz
★
1.9k
We have a data set from Drosophila melanogaster, we don't know much about. We ran the Seurat-workflow and got various clusters.
I would like to try and annotate this clusters in an automated manner.
I know the SingleR package is the one to go with.
I would like to know though, if there are other tools to run cell-type annotations for single-sell data sets.
Also, As we are working with D. melanogaster, I would like to find appropriate references to work with.
Looking at the celldex package, it is mainly for human (and a bit for mouse). Can someone tell me where and if I can find a reference data set for this organism?
Does someone know of an available reference set for this organism?
thanks in advance
Assa
What's the tissue/cell type you are working with?
There is flyCellAtlas for drosophila single cell during different developmental stages. You can this as a reference and perform reference mapping for cell type annotation. However you need to be careful if you are using some treatment or perturbed condition.
Just out of curiosity, is there any specific reason for using automated cell type annotation? Because generally in scRNA-Seq I prefer finding cluster specific DEGs and select the markers (based on fold change and their specificity) for cell type annotation.
Regards,
Nitin N.
We're working on egg chambers. in different developmental stages
We did go the manual way, but for some of the clusters we have difficulties assigning DEG.
I'll look into the atlas. thanks.
This happens if you have different state for same type of cells. Let's say if cells are undergoing differentiation, it would be really difficult to get specific set of differential expressed genes for such clusters/subgroup.
Better way to deal with this is to build a trajectory and try to recapitulate the changes in gene expression over the trajectory. Arranging genes over pseudotime in your trajectory based on their expression might be useful to detect the changes in different cell states.
All the best.
Regards,
Nitin N.
thanks.
Do you have any experience with the fly cell atlas?
I was looking at the five data sets from the ovaries. Are they all one data set, analyzed with different technologies/stringency?
No, I have never used this data in any of my project.
As far as I know (based on their publication) they have sequenced the samples both on 10X (droplet-based) and SMART-Seq2 (Well-based) platforms.
About stringency, you can find information here on their website (https://flycellatlas.org/#tutorials), FAQ section. Briefly, in stringent analysis they have used ambient RNA-correction step with some additional filters.
Regards,
Nitin N.