Alignment

Clusters from each sample are aligned against amplicon sequences specified in the manifest file.

Each paired-end read is initially evaluated in terms of its alignment to the relevant probe sequences for that read. Read 1 is evaluated against the reverse complement of the Downstream Locus-Specific Oligos (DLSO) and Read 2 is evaluated against the Upstream Locus-Specific Oligos (ULSO). If the start of a read matches a probe sequence with no more than one mismatch, the full length of the read is aligned against the amplicon target for that sequence. This alignment is performed along the length of the amplicon target using a banded Smith-Waterman alignment algorithm.

The banded Smith-Waterman algorithm performs local sequence alignments to determine similar regions between two sequences. Instead of looking at the total sequence, the Smith-Waterman algorithm compares segments of all possible lengths, given the restriction that the maximum indel size is 25 bp.

Any alignments that include more than three indels are filtered from alignment results. Filtered alignments are written in alignment (BAM) files as unaligned and are not used in variant calling. Indels within the DLSO and ULSO are not observed given the assay chemistry.