flowchart TB A[From cluster]--Select 1st consistent marker\nand other clusters markerd by it--> C[Find distinct genes within] B[From gene]--Use it as the 1st marker\nand find all clusters consistently marked by it--> C C--Find genes that mark only one\ncluster among the above ones--> D[Co-expression plot] D--Select the 2nd marker and\nsee the expression of the marker combo\nin the whole dataset--> E[Plot expression trend]
4 Recommended Workflow
scMarko aims to help with the heuristic selection of markers for your cell types of interest by:
4.1 Recommended Workflow
4.2 Select your first marker
Depends on the prior knowledge about the diversity in your tissue of interest, you might start from a cell type (From cluster) and select a gene that marks it. scMarko will provide a list of other clusters (offtargets) that the second marker in the combination aims to distinguish.
Sometimes, you are interested in cell types that are marked by the same gene, and that gene will be your first marker. scMarko will list all clusters that are likely to express this gene so you can find putative second markers that will further tell them apart (From gene).
4.3 Select a second marker
Find distinct genes lists the genes that are expressed in only one of the clusters from From cluster or From gene. We hope the clusters that you are interested in have at least one such gene. If not, you could select the clusters that co-express your first two markers, and run Find distinct genes on them to find a third marker that distinguish them.
4.4 Inspect your marker combo
While the marker combinations selected by this process should be specific among the dataset, there could be gray zones during the binarization process. To see if this is the case, plotting normalized expression value and see how your clusters of interest compare to others on this scale could provide some insight.
Plot expression trend plots normalized expression values as a line plot across stages/conditions and highlights the cluster you selected.
To inspect how your marker combinations look like in your dataset (e.g., which clusters expresses either of them at a stage/condition), Co-expression plot plots this information for your combinations of markers.