mappability

Mappability is a measure of how uniquely a genomic position can be identified given sequencing reads and a reference genome. For a fixed read length, it describes the likelihood that a read originating at a particular location maps back to that exact location rather than to other similar sites. In practice, mappability is often expressed as the fraction of positions in the genome that can be uniquely mapped under a specified alignment criterion, such as the number of allowed mismatches.

Mappability can be assessed as k-mer mappability, where k equals the read length. A position is considered uniquely mappable if the corresponding k-mer occurs only once in the reference under the chosen criteria. Conversely, multi-mappable regions contain sequences that occur in multiple genomic locations, typically due to repeats, segmental duplications, or gene families. Mappability is not absolute; it depends on read length, sequencing error rates, and the alignment model used.

Several factors influence mappability, including repetitive DNA, paralogous regions, genome assembly gaps, and reference genome quality. Longer reads generally increase mappability, while higher error tolerance decreases it. Mappability maps are used to inform downstream analyses by identifying regions where read-based inferences (such as variant calling or expression quantification) may be unreliable and by guiding the design of experiments or analytical pipelines.

Common applications include masking low-mappability regions in genome analyses, evaluating alignment strategies, and interpreting results from sequencing experiments. Tools and datasets, such as mappability tracks and specialized mappers, help compute or approximate mappability for given read lengths and alignment settings. Examples of related tools include GenMap and mappability resources like Umap/Bismap.