This may sound like a vague question but your help and inputs on this could really help me move forward. I am helping a bench scientist with his single cell RNA sequencing data generated using drop-seq platform. Memory cells CCR7+/CD45RA- were used for this experiment and were extracted from blood.I did the basic QC analysis and everything seems to be fine with that. Then I performed clustering to find genes with similar expression profile, I did that by using seurat(http://satijalab.org/seurat/pbmc-tutorial.html). I'm almost done with the analysis and my t-SNE plot shows me 5 decent clusters. I guess at the end I would have to label each cluster by a cell type provided that a particular cell-type is known to express a marker-gene which might be a part of a cluster. Is there a universal list of cell-type gene marker that is available and could be applied with what I am doing? Please let me know if the question makes sense, I am trying to do something like the "Assigning cell type identity to clusters" section of the seurat tutorial(http://satijalab.org/seurat/pbmc-tutorial.html).
Memory cells CCR7+/CD45RA- were used in the experiment and were extracted from blood.
Seurat is not a Bioconductor package, you're better off asking the package authors directly.
But for what it's worth: if I had a good list of cell type-specific markers that could be applied reliably in any context, I wouldn't need to collaborate with biologists anymore to interpret my analysis results. I could just pay a technician to generate scRNA-seq data for me, shove it through my pipelines, get the computer to assign a cell type to each cluster, and rinse and repeat until I get enough data for a Nature/Science/Cell paper.
In summary, the interpretation of the gene list is where the biologists earn their pay. Generally speaking, they do know what they're doing, and it would be a shame to let that knowledge go to waste. I think of them as neural networks that have been trained for 5-10 years on identifying features of interest for a particular biological system; that would probably match any of the deep learning algorithms out there today.
Hrishi, much of the power of single cell RNA-seq lies in the ability to detect rare or novel populations, and much of the literature on markers is often limited because of the small number of markers people were looking at in prior studies or the abundance of these populations in bulk data. No pre-defined list of markers is necessarily good enough to describe any and all possible populations, but one way of seeing if your categories overlap with previously well-defined populations is to pick marker genes using FindAllMarkers or FindMarkers in Seurat and then chuck that into a pathway analysis or gene ontology analysis package.
