ProPhex

ProPhex is an efficient k-mer index with a small memory footprint. It uses the BWA implementation of the BWT-index. ProPhex is designed as a core computational component of ProPhyle, a phylogeny-based metagenomic classifier allowing fast and accurate read assignment.

Getting started

git clone https://github.com/prophyle/prophex
cd prophex && make -j

Alternative ways of installation

conda install prophex

Quick example

# Build a ProPhex index
./prophex index -k 25 index.fa

# Query reads from reads.fq for k=25 (with k-LCP)
./prophex query -k 25 -u -t 4 index.fa index.fq

# Query reads from reads.fq for k=20 (with 4 threads and without k-LCP)
./prophex query -k 20 index.fa index.fq

ProPhex commands

Program: prophex (a lossless k-mer index)
Version: 0.1.1
Authors: Kamil Salikhov, Karel Brinda, Simone Pignotti, Gregory Kucherov
Contact: kamil.salikhov@univ-mlv.fr

Usage:   prophex <command> [options]

Command: index           construct a BWA index and k-LCP
         query           query reads against index

         klcp            construct an additional k-LCP
         bwtdowngrade    downgrade .bwt to the old, more compact format without Occ
         bwt2fa          reconstruct FASTA from BWT

Usage:   prophex index [options] <idxbase>
Options: -k INT    k-mer length for k-LCP
         -s        construct k-LCP and SA in parallel
         -i        sampling distance for SA
         -h        print help message

Usage:   prophex query [options] <idxbase> <in.fq>

Options: -k INT    length of k-mer
         -u        use k-LCP for querying
         -v        output set of chromosomes for every k-mer
         -p        do not check whether k-mer is on border of two contigs, and show such k-mers in output
         -b        print sequences and base qualities
         -l STR    log file name to output statistics
         -t INT    number of threads [1]
         -h        print help message

Usage:   prophex klcp [options] <idxbase>

Options: -k INT    length of k-mer
         -s        construct k-LCP and SA in parallel
         -i        sampling distance for SA
         -h        print help message

Usage:   prophex bwtdowngrade <input.bwt> <output.bwt>
         -h        print help message

Usage:   prophex bwt2fa <idxbase> <output.fa>
         -h        print help message

Output format

Matches are reported in an extended Kraken format. ProPhex produces a tab-delimited file with the following columns:

Category (unused, U as a legacy value)
Sequence name
Final decision (unused, 0 as a legacy value)
Sequence length
Assigned k-mers. Space-delimited list of k-mer blocks with the same assignments. The list is of the following format: comma-delimited list of sets (or A for ambiguous, or 0 for no matches), colon, length. Example: 2157,393595:1 393595:1 0:16 (the first k-mer assigned to the nodes 2157 and 393595, the second k-mer assigned to 393595, the subsequent 16 k-mers unassigned)
Bases (optional)
Base qualities (optional)

FAQs

Can I remove duplicate k-mers from the index in order to use less memory when querying?

Yes, duplicate k-mers can be removed using ProphAsm, which assembles contigs by greedy enumeration of disjoint paths in the associated de-Bruijn graph. BCalm is another tool that can be used with ProPhex. Compared to ProPhex, BCalm has a smaller memory footprint. On the other hand, the resulting FASTA file can be significantly bigger (when assemblying, BCalm stops at every branching k-mer).