I have two cell-by-gene matrices, each representing gene counts for cells in sample 1 and sample 2. I'm interested in identifying common gene expression patterns across both samples. These patterns consist of gene sets that are interdependent and likely contribute to specific biological processes.
Some methods use variational autoencoders (VAEs) to detect pre-defined gene expression patterns within each dataset, comparing potential biological processes afterward (this paper). Considering that VAEs model the distribution of latent factors, could they be used to find similar gene expression patterns across the two datasets? I do not have pre-defined patterns and aim to find similarities in an unsupervised way. Or would it be more effective to compare the latent factors derived from standard autoencoders? I’m looking for guidance on the best approach to take.
