Empirical distribution of k-word matches in biological sequences

This study focuses on an alignment-free sequence comparison method: the number of words of length k shared between two sequences, also known as the D_2 statistic. The advantages of the use of this statistic over alignment-based methods are firstly that it does not assume that homologous segments are contiguous, and secondly that the algorithm is computationally extremely fast, the runtime being proportional to the size of the sequence under scrutiny. Existing applications of the D_2 statistic include the clustering of related sequences in large EST databases such as the STACK database. Such applications have typically relied on heuristics without any statistical basis. Rigorous statistical characterisations of the distribution of D_2 have subsequently been undertaken, but have focussed on the distribution’s asymptotic behaviour, leaving the distribution of D_2 uncharacterised for most practical cases. The work presented here bridges these two worlds to give usable approximations of the distribution of D_2 for ranges of parameters most frequently encountered in the study of biological sequences.

💡 Research Summary

This paper investigates the practical distribution of the D₂ statistic, which counts the number of exact k‑word matches between two biological sequences, and provides usable approximations for the range of parameters most commonly encountered in bioinformatics. The D₂ statistic is attractive because it is alignment‑free, assumes no contiguity of homologous regions, and can be computed in linear time with respect to sequence length. While previous work has derived asymptotic results (normal distribution for relatively short k, compound Poisson for large k), those results do not describe the distribution for finite, biologically realistic sequence lengths (hundreds to a few thousand bases or residues).

The authors first formalize the problem: two sequences A (length m) and B (length n) are modeled as independent identically distributed (i.i.d.) Bernoulli texts over an alphabet 𝔸 (DNA or amino‑acid). A binary indicator Yᵢⱼ equals 1 when the k‑mer starting at position i in A matches the k‑mer starting at position j in B; D₂ is the sum of all Yᵢⱼ over the appropriate index set. The expected value is E