spacedkmer

Spaced k-mer is a representation used in sequence analysis in which a k-mer is defined by a fixed pattern of positions rather than by a contiguous run of bases. In this approach, a mask or pattern specifies which positions within a window of length L are considered for matching. The number of positions kept by the pattern is k, and the resulting subsequence formed from those selected positions is the spaced k-mer. For example, with a mask 1-0-1-1-0-1 on a six-base window, the bases at positions 1, 3, 4, and 6 are used, yielding a spaced k-mer of those four bases from the corresponding window.

Spaced k-mers are part of the broader concept of spaced seeds, and they are used to improve sensitivity in sequence comparison, read mapping, and assembly. By excluding certain positions, especially those prone to errors or variability, spaced k-mers can detect similarities that contiguous k-mers might miss, particularly in the presence of substitutions or indels. They are often employed in seed-based search algorithms to identify candidate alignments that are then refined by more precise alignment.

Construction and analysis typically involve hashing or indexing the selected positions defined by the mask, enabling fast lookup of potential matches across large genomic datasets. The choice of pattern influences sensitivity and specificity, and researchers may use multiple masks or optimize masks for particular error profiles or evolutionary distances.

Limitations include the need for careful pattern design and potential increases in computational overhead or complex indexing compared to using standard contiguous k-mers. Nonetheless, spaced k-mers have been shown to enhance detection power in various genomic applications and are a foundational concept in seed-based sequence analysis.