TMAP alignment refinement

The TMAP alignment refinement phase is organized as a pipeline of optional processing stages, each controlled by a specific set of options. The following stages are available (given in the order of their optional invocation).

Table 1.

Stage

Description

Flow space realignment

Performs alignment of the sequence of read flow signals to simulated reference flow signal. Turned on with the --final flowspace flag.

Context-based realignment

Realigns read with reduced cost of INDELs within homopolymers. Turned on with the --context flag.

Local realignment with reduced gap cost

Realigns specific zones that are likely to contain errors associated with signal phasing using reduced gap cost. This helps to reduce false positive variants in low complexity zones. Turned on with the --do-realign flag.

Long INDEL salvage

Realigns zones adjacent to the read ends while allowing for longer indels. Helps to find long indels continued beyond the ends of the amplicons. Turned on by specifying long INDEL gap penalty using the (-G option.

End repair

Extends the alignment beyond the amplicon edges, where suitable. Also trims alignment tails below given mismatch ratio. This is a simplified and often better performing version of the long indel salvage option. It is turned on by specifying the --end-repair <MM> option on the TMAP command line, where <MM> is the maximum allowed mismatch percentage at the alignment edge. End repair is typically used together with the --J option (max adapter bases for soft clipping), where recommended usage is --end-repair 25-J 15).

Bed file

The BED file specification, given with the --bed-file file_name option, provides TMAP with the amplicon boundary coordinates. This is used by end-repair to extend gaps over amplicon edges. It also alters the way mapping scores are calculated: the reads mapped within amplicon boundaries are given a 12-point boot in MAPQ, and thus are preferred over the alternative mapping locations outside of the amplicon set.

5' softclip removal

Removes soft clips from the 5' end of the alignment if introduced by end-repair. This option is on by default. Turned off with the --er-5clip flag.

3' tandem repeat clipping

Clips tandem repeats from 3' end of the reads. Helps reduce variant detection errors in tandem repeat zones. Turned on with the --do-repeat-clip flag.