With recent advancements in single cell sequencing technologies, routine data analysis activities include identifying cell subtypes in tissues and understanding their relationships at the single cell level. While existing algorithms excel in distinguishing different cell types based on known markers, subtyping cells based on their functions remains to be a challenge. To address this limitation, we propose a new single cell subtyping method, called Abstract Functional Analysis (AFA), which incorporates a priori known context-specific biological processes into the analysis. The key premise of AFA is that interjecting “some form of prior knowledge” (GO Biological Processes in our case) into the otherwise unbiased analysis is amenable to deriving, namely, biological function centric subtyping. We assessed our method on eight publicly available Alzheimer’s Disease related single-cell mRNA datasets and demonstrated that AFA can subgroup cells based on their functional roles into subtypes such as disease associated microglia (DAM) and late-stage homeostatic microglia exhibiting DAM signature. Advantages of AFA include labeling subtypes based on functions and discovering additional biological processes enriched within each identified subtype. AFA offers a new way of subgrouping and naming cells thereby enhancing our understanding of cellular heterogeneity in a more intuitive and useful way.