scMAPA

Alternative polyadenylation (APA) causes shortening or lengthening of the 3ʹ-untranslated region (3ʹ-UTR), widespread in complex tissues. To detect APA and identify cell-type-specific APA in a multi-cluster setting, we developed a model-based method, scMAPA. The first part of scMAPA is coded as shell scripts, which can 1) divide the aligned read data by the cell cluster[1] and remove PCR duplicates by UMI-tools; 2) Pad the 3'biased reads and convert BAM to Bedgraph file; 3) estimate the abundance of 3ʹ-UTR long and short isoform of genes in each cluster-bulk data using linear regression and quadratic programming implemented in DaPars2. The second part of scMAPA is coded as an R package, which can 4) fit a logistic regression model for each gene and estimate the significance of APA; 5) Identify cluster-specific 3'UTR shortening and lengthening; 6) Do visualization to show the APA dynamics.