Why monocytes have some cells together with neutrophil? Cells belong to the same cluster but position with other cluster. How should i face the problem.
Your question omits a lot of details which will make a big difference on the interpretation of this data. Lets assume that you've followed something like the Seurat vignette, and nothing more special than that, ie batch correction. You've annotated two sets of cells as Neutrophils and Monocytes, and you're wondering why they're not discrete, why does one bleed into the other?
Firstly, while both cells originate from the myeloid lineage, they should be fairly distinct in lower dimensional space. Have you looked at specific markers such as CD66b (CEACAM8) for neutrophils, or CD14/16 for monocytes?
From a more technical standpoint, there are a few variables which contribute to the picture you've created; number of highly variable genes (and method), number of principal components, clustering method (and parameters, such as resolution). Each of these will contribute to the image you see above. The fact that there's overlap between your labelled monocytes and neutrophil populations can also be because we're projecting multidimensional data onto two planes, there is often bleed through, but that's because there's shared variance between the two populations. If the variance shared is a concern, then I'd encourage you to look at your HVG method and parameters.
There's a lot of contention in the field as to how much stock people put in two dimensional embeddings relative to the technical and biological complexity they're trying to capture. More often than not figures such as UMAP are over interpreted and this progresses to misleading conclusions / hypotheses which then struggle to validate.
It's good to take a step back at this point and understand what your goal is with the analysis.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Your question lacks any details of what we see here, how you identified clusters and how you annotated.
Sorry, I shouldn't have used this example, but in simple terms, monocytes come from the same cluster, but some of the cells in this cluster are in the same position as the neutrophil cells.
I would argue that is poor annotation then or choice of variable genes and/or clustering resolution is not good.
Thanks for you answers, I prefer that this is a discrepancy between the FindClusters and RunMAP methods.