Fig. 2

Pipeline for immunoglobulin gene reconstruction in human samples. The pipeline used for IgH and IgL gene reconstruction using either all sequencing reads (Unfiltered) or bioinformatically filtered reads (IG_mapped, IG_mapped+Unmapped, Recombinome_mapped, and IMGT_mapped) from sc-RNA-seq data. Details for each filter are described in Methods and in the text. In the initial step, adapter sequences are trimmed from the fastq files using Trimmomatic. Reads are then filtered to enrich those containing partial sequences from the IgH or IgL variable region and constant regions, and to exclude reads mapping to conventional protein coding genes. Filtered (or total) reads are then assembled using the Trinity algorithm without normalization. The assembled transcript models are annotated using IgBLAST. The reads used for assembly are mapped to the assembled transcript models using bowtie2. The models are ranked according to the number of reads mapped. Transcript models that are not productive or have a V(D)J and CDR nucleotide sequence that is the same as a higher ranked model are filtered out. The top model from the remaining set is selected as the putative heavy or light chain