kmindex

kmindex is a tool for indexing and querying sequencing samples. It is built on top of kmtricks.

Given a databank $D=\{S_1,\mathrm{.}\mathrm{.}\mathrm{.},S_n\}D\; =\; \backslash \{S\_1,\; ...,\; S\_n\backslash \}$, with each $S_{i}S\_i$ being any genomic dataset (genome or raw reads), kmindex allows to compute the percentage of shared k-mers between a query $QQ$ and each $S\in DS\; \backslash in\; D$. It supports multiple datasets and allows searching for each sub-index $D_i\in G=\{D_1,\mathrm{.}\mathrm{.}\mathrm{.},D_m\}D\_i\; \backslash in\; G\; =\; \backslash \{D\_1,...,D\_m\backslash \}$. Queries benefit from the findere algorithm. In a few words, findere allows to reduce the false positive rate at query time by querying $(s+z)(s+z)$-mers instead of $ss$-mers, which are the indexed words, usually called $kk$-mers. kmindex is a tool for querying sequencing samples indexed using kmtricks.

Indexing/Querying example (can be tested in the examples directoy):

1. Index a dataset:

   kmindex build --fof fof1.txt --run-dir D1_index --index ./G --register-as D --hard-min 2 --kmer-size 25 --nb-cell 1000000

2. Query the index:

   kmindex query --index ./G --fastx query.fasta --zvalue 3

Full documentation is available at https://tlemane.github.io/kmindex

Citation Lemane, Téo, et al. “Indexing and real-time user-friendly queries in terabyte-sized complex genomic datasets with kmindex and ORA“ Nature Computational Science 4.2 (2024): 104-109.

Pre-print paper is available on bioRxiv