Generalized Gapped-kmer Filters for Robust Frequency Estimation.

In this paper, we study the generalized gapped k-mer filters and derive a closed form solution for their coefficients. We consider nonnegative integers $\ell$ and $k$, with $k\leq \ell$, and an $\ell$-tuple $B=(b_1,\ldots,b_{\ell})$ of integers $b_i\geq 2$, $i=1,\ldots,\ell$. We introduce and study an incidence matrix $A=A_{\ell,k;B}$. We develop a Mobius-like function $\nu_B$ which helps us to obtain closed forms for a complete set of mutually orthogonal eigenvectors of $A^{\top} A$ as well as a complete set of mutually orthogonal eigenvectors of $AA^{\top}$ corresponding to nonzero eigenvalues. The reduced singular value decomposition of $A$ and combinatorial interpretations for the nullity and rank of $A$, are among the consequences of this approach. We then combine the obtained formulas, some results from linear algebra, and combinatorial identities of elementary symmetric functions and $\nu_B$, to provide the entries of the Moore-Penrose pseudo-inverse matrix $A^{+}$ and the Gapped k-mer filter matrix $A^{+} A$.