How subsetting data and subclusterıng affect the results?

[kiwipeel]

I want to analyze only a specific cell type in my data so ı subsetted my data. I did standart seurat workflow and after clusterıng ı realized that some clusters contains compact subclusters. I came to conclusion that these small subclusters (100-200 cell , total subsetted data contains 8000 ) are real subtypes by checking markers in literature. Diving clusters to subclusters, would this affect the results? I can also find more subclusters since cells in my data is hard to annotatate and mostly in trasition state. Or should ı use major clusters for propeller function?