NIH Research Festival
–
Single cell and spatial omic technologies continue to inundate the field of neurobiology with increasingly complex data investigating cellular states and responses. Each dataset produced provides more context into how a particular cell reacts to genetic manipulation, the presence of disease-relevant stimuli, and the influence of environmental factors. However, the field lacks a singular consensus methodology for cell classification that is robust enough to withstand the large batch-, laboratory-, and platform-specific noise present in single cell and spatial data. Although many computationally intensive pipelines are under development, there is a need to streamline iterative gene-by-cluster annotation strategies. We propose a visualization-forward approach to cell classification that we call semi-automated hand annotation (SAHA). By creating an easily accessible package in R, we have lowered the barrier for entry-level and skilled bioinformaticians alike to 1) compare the transcriptional state of clusters within their own data, 2) identify overlap between their own clusters and any publicly available database, and 3) implement marker-free cell similarity analysis to overcome challenges in targeted or low-sensitivity experiments. SAHA was specially designed to be able to run on a desktop or laptop computer, where only summary files of a dataset may be accessible. SAHA provides the user with a classification, summaries of how that classification was reached, and the flexibility to choose a marker database that is most appropriate for their study. In contrast to machine-learning or integration-based annotation approaches that are fully automatic, SAHA’s semi-automatic pipeline allows researchers several checkpoints to fine-tune their annotations.
Scientific Focus Area: Genetics and Genomics
This page was last updated on Tuesday, August 6, 2024