mapquik is an ultra-fast read mapper based on $k$-min-mers (matches of $k$ consecutively-sampled minimizers). It aligns long and accurate reads such as PacBio HiFi to a reference genome.

The underlying seed constructs ($k$-mers) in state-of-the-art long-read mappers are tailored to noisy reads, and small seed sizes induce longer computation times due to multiple potential mapping locations. Recent advances in short-read alignment methods have demonstrated that 98% of many organisms' genomes are non-repetitive and can be uniquely aligned to with longer seeds. Therefore, we explore the use of longer, non-exact seeds ($k$-min-mers) in accurate long reads. See our manuscript for details.

The mapping performance of mapquik degrades markedly when identity between reads and the reference is lower than $97$%, and less than $1$% of the reads are mapped at $Q60$ for identities below $93$%. Therefore, mapquik is not suitable for mapping PacBio CLR reads, and potentially also Oxford Nanopore reads until base-calling consistently reaches identity levels above $98$%.

Pre-requisites: A working Rust environment.

Clone the repository, and run

rustup install nightly
cargo +nightly build --release

The nightly version of cargo is required because mapquik uses experimental language features (such as SIMD and intrinsics).

target/release/mapquik <reads.fq> --reference <reference.fa>

mapquik takes a single FASTA/FASTQ input (gzip-compressed or not) as input. Multi-line sequences are not supported.

The output of mapquik is a regular PAF file.

An example reference genome, and a script to simulate reads using pbsim are provided in the example/ folder. To run mapquik on a small set of 100 reads, type:

cd example && bash run_ecoli.sh

which will run both mapquik and minimap2 on 100 simulated reads, and return the output of paftools.js mapeval on both PAF files.

To simulate a larger set of reads using pbsim and map, type:

bash simulate_pbsim.sh && bash run_ecoli_full.sh

For further information on usage and parameters, run

target/release/mapquik -h

for a one-line summary of each flag, or run

target/release/mapquik --help

for a lengthy explanation of each flag.

All scripts used to generate the figures and tables in the paper can be found in the experiments/ folder. Specifically, the simulate_chm13.sh and simulate_maize.sh scripts can be used similarly to simulate reads.

In order to obtain and map DeepConsensus reads, first run

wget https://storage.googleapis.com/brain-genomics-public/research/deepconsensus/data/v0.3/assembly_analysis/fastqs/HG002_24kb_2SMRT_cells.dc.v0.3.q20.fastq.gz
gunzip -c HG002_24kb_2SMRT_cells.dc.v0.3.q20.fastq.gz | grep -v TOTAL > dc.hg002.fastq

and map to a reference genome reference.fa in your directory with mapquik using

target/release/mapquik dc.hg002.fastq --reference reference.fa -p mapquik-dc

mapquik significantly accelerates the seeding and chaining steps for both the human and maize genomes with $>96$% sensitivity and near-perfect specificity. On the human genome, for both real and simulated reads, mapquik achieves a $37\times$ speed-up over minimap2, and on the maize genome, a $410\times$ speed-up over minimap2.

mapquik indexing is $9\times$ faster than minimap2, which is of independent interest.

mapquik is freely available under the MIT License.

• Barış Ekim, supervised by Bonnie Berger at the Computer Science and Artificial Intelligence Laboratory (CSAIL) at Massachusetts Institute of Technology (MIT)
• Rayan Chikhi at the Department of Computational Biology at Institut Pasteur

@article{mapquik,
  title={Efficient mapping of accurate long reads in minimizer space with mapquik},
  author={Ekim, Bar{\i}{\c{s}} and Sahlin, Kristoffer and Medvedev, Paul and Berger, Bonnie and Chikhi, Rayan},
  journal={Genome Research},
  pages={gr--277679},
  year={2023},
  publisher={Cold Spring Harbor Laboratory}
}

Should you have any inquiries, please contact Barış Ekim at baris [at] mit [dot] edu, or Rayan Chikhi at rchikhi [at] pasteur [dot] fr.