Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Taxonomic classification with maximal exact matches in KATKA kernels and minimizer digests

Dominika Draesslerová, Omar Ahmed, Travis Gagie, Jan Holub, Ben Langmead, Giovanni Manzini, Gonzalo Navarro

TL;DR

This work addresses MEM-based taxonomic classification in large, repetitive genome collections by proposing lossy representations—KATKA kernels, minimizer digests, and KATKA kernels of minimizer digests—for building compact augmented FM-indexes that support MEM queries. The authors formalize augmented FM-indexes and three kernelized string representations, and demonstrate how MEM tables can be approximated under these representations with favorable space-time tradeoffs. Empirical results on 1000 bacterial genera show substantial compression (e.g., tens to hundreds of MiB savings) with modest declines in true-positive MEM mapping rates and potential speedups, especially for kernelized digests. Overall, the KATKA-kernel-of-minimizer-digest approach appears to combine the benefits of kernelization and minimization to enable scalable MEM-based classification on expanding reference databases.

Abstract

For taxonomic classification, we are asked to index the genomes in a phylogenetic tree such that later, given a DNA read, we can quickly choose a small subtree likely to contain the genome from which that read was drawn. Although popular classifiers such as Kraken use $k$-mers, recent research indicates that using maximal exact matches (MEMs) can lead to better classifications. For example, we can build an augmented FM-index over the the genomes in the tree concatenated in left-to-right order; for each MEM in a read, find the interval in the suffix array containing the starting positions of that MEM's occurrences in those genomes; find the minimum and maximum values stored in that interval; take the lowest common ancestor (LCA) of the genomes containing the characters at those positions. This solution is practical, however, only when the total size of the genomes in the tree is fairly small. In this paper we consider applying the same solution to three lossily compressed representations of the genomes' concatenation: a KATKA kernel, which discards characters that are not in the first or last occurrence of any $k_{\max}$-tuple, for a parameter $k_{\max}$; a minimizer digest; a KATKA kernel of a minimizer digest. With a test dataset and these three representations of it, simulated reads and various parameter settings, we checked how many reads' longest MEMs occurred only in the sequences from which those reads were generated ("true positive" reads). For some parameter settings we achieved significant compression while only slightly decreasing the true-positive rate.

Taxonomic classification with maximal exact matches in KATKA kernels and minimizer digests

TL;DR

This work addresses MEM-based taxonomic classification in large, repetitive genome collections by proposing lossy representations—KATKA kernels, minimizer digests, and KATKA kernels of minimizer digests—for building compact augmented FM-indexes that support MEM queries. The authors formalize augmented FM-indexes and three kernelized string representations, and demonstrate how MEM tables can be approximated under these representations with favorable space-time tradeoffs. Empirical results on 1000 bacterial genera show substantial compression (e.g., tens to hundreds of MiB savings) with modest declines in true-positive MEM mapping rates and potential speedups, especially for kernelized digests. Overall, the KATKA-kernel-of-minimizer-digest approach appears to combine the benefits of kernelization and minimization to enable scalable MEM-based classification on expanding reference databases.

Abstract

For taxonomic classification, we are asked to index the genomes in a phylogenetic tree such that later, given a DNA read, we can quickly choose a small subtree likely to contain the genome from which that read was drawn. Although popular classifiers such as Kraken use -mers, recent research indicates that using maximal exact matches (MEMs) can lead to better classifications. For example, we can build an augmented FM-index over the the genomes in the tree concatenated in left-to-right order; for each MEM in a read, find the interval in the suffix array containing the starting positions of that MEM's occurrences in those genomes; find the minimum and maximum values stored in that interval; take the lowest common ancestor (LCA) of the genomes containing the characters at those positions. This solution is practical, however, only when the total size of the genomes in the tree is fairly small. In this paper we consider applying the same solution to three lossily compressed representations of the genomes' concatenation: a KATKA kernel, which discards characters that are not in the first or last occurrence of any -tuple, for a parameter ; a minimizer digest; a KATKA kernel of a minimizer digest. With a test dataset and these three representations of it, simulated reads and various parameter settings, we checked how many reads' longest MEMs occurred only in the sequences from which those reads were generated ("true positive" reads). For some parameter settings we achieved significant compression while only slightly decreasing the true-positive rate.
Paper Structure (9 sections, 11 figures)

This paper contains 9 sections, 11 figures.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Augmented FM-indexes
  4. String kernels and KATKA kernels
  5. Minimizer digests
  6. KATKA kernels of minimizer digests
  7. Approximating MEM tables with FM-indexes of KATKA kernels and minimizer digests
  8. Experiments
  9. Conclusions and future work

Figures (11)

  • Figure 1: A toy phylogenetic tree (top) with Kraken's $k$-mer index for $k = 3$(bottom).
  • Figure 2: The augmented FM-index for our toy collection of genomes.
  • Figure 3: A slightly larger collection of slightly longer toy genomes.
  • Figure 4: The subsequence consisting of the characters in the first or last occurrence of each distinct 4-mer and the copies of $, with omitted characters replaced by spaces.
  • Figure 5: The subsequence consisting of the characters in the first or last occurrence of each distinct 4-mer --- the 4th-order KATKA kernel --- and the copies of $, with maximal omitted substrings replaced by copies of #, except for those adjacent to $.
  • ...and 6 more figures