High-sensitivity mapping
High-sensitivity mapping is recommended when target references are similar or when low-abundance references must be detected. In these cases, the standard mapping can split the input reads across similar target references and therefore underestimate their abundance. The high-sensitivity mapping therefore collects broader evidence for the abundance of candidate target references before calculating the final mapping statistics.
The following steps are performed:
- In a first mapping step, input reads are mapped to the target references and, optionally, the host references, using the chosen Non-specific match handling option. Reads mapping equally well to multiple target references, or to multiple host references, are assigned randomly.
- All reads mapping to a target reference in the first step, referred to as target reads, are then mapped again to all target references. In this step, reads mapping equally well to multiple target references can be assigned to up to 16 references. As a result, the same read can contribute support to more than one reference, which helps similar target references receive sufficient evidence instead of forcing a single random choice.
- Target references with at least one mapped read are selected as candidate target references.
- For performance reasons, regions with coverage above the configured Maximum coverage threshold are downsampled to obtain a coverage of at most that threshold. Very high coverage often indicates insufficient read trimming or many reads originating from low-complexity regions.
- The target reads, after the optional downsampling, are mapped to the candidate target references using BLAST.
- Mapping statistics for the candidate target references are then calculated from the BLAST hits that satisfy the Minimum percent identity (%) and Minimum overlap (%) thresholds.