Identification of the adapter, barcode, UMI, and poly(A)-tail sequences from Nanopore reads

We removed the cDNA sequences from Nanopore reads and extracted up to 100 bp from both ends. We developed a modified version of FLEXBAR (Dodt et al. 2012; Roehr et al. 2017) to align P1 primer adapter sequence with the following parameters (“-ao 10 -ae 0.3 -ag -2 -hr T -hi 10 -he 0.3 -be 0.2 -bg -2 -bo 5 -ul 26 -kb 3 -fl 100”). Then we aligned the Nanopore reads that have valid adapters to the cellular barcodes which have been previously identified by Cell Ranger. We scanned the poly(A) sequences using the homopolymer-trimming function of FLEXBAR downstream from the cell barcode. Once the poly(A) sequences were found, the UMI sequences between the poly(A) and barcode were searched using MUMmer 4.0 (Marçais et al. 2018) (with parameters “-maxmatch -b -c -l 7 -F”) and in-house scripts against the Illumina UMIs of the same cell and the same gene or genomic regions (±500 bp from each end of the reads). In the end, ScNapBar output the alignment score of the adapter, the number of mismatches and indel from the barcode alignment, the length of poly(A) and UMI sequences, as well as the length of the gap between the barcode and adapter. We use these features to estimate the likelihood of the barcode assignment in the steps shown in Figure 1.