MEM Mapper Prototype: mapping short reads, faithfully

---

Contents:

1. Prototype? What do you mean?
3. Compilation and installation.
4. Running mmp

---

I. Prototype? What do you mean?

mmp is a proper mapper, we use it in the lab and you are more than welcome to use it as well. When we say ‘prototype’, we mean that our purpose was not really to secure a stable user base, but rather to show that faithful mapping is achievable.

Below are the main things to consider before using mmp:

1. It works only with single-end reads.
2. The whole read is aligned, there is no trimming.
3. There is no way to change the sensitivity.

Otherwise, mmp is pretty good. The most important feature is that it is approximately faithful, so you should be able to trust the mapping quality. You will notice that some reads have a mapping quality over 150. It is not a bug, we mean it: the location of those reads is extremely likely to be correct.

Finally, mmp is quite fast but it uses a little more memory than many mainstream mappers.

You can find more information about mmp in the bioRxiv preprint that describes the mapping strategy.

II. Compilation and installation

To install mmp, open a terminal and clone this git repository:

git clone https://github.com/gui11aume/mmp

The files should be downloaded in a directory named mmp. Go to this new directory and use make to compile:

cd mmp
make

This will create a binary file mmp.

III. Running mmp

mmp runs on Linux.

Indexing:

You first have to index a genome file before you can map reads in it. If the genome of interest is saved in a file called genome.fasta, you index it with the following command:

./mmp --index genome.fasta

This creates a bunch of files that all start with genome.fasta.. This is the index.

Mapping:

Once the genome is indexed, you can start mapping reads in it. For this you must know the approximate error rate of the sequencer. By default, mmp assumes that it is 1%.

If your reads are in a file called reads.fastq, you map them with the following command:

./mmp genome.fasta reads.fastq

If you want to use more than one core to speed up the mapping, you can use the option -t. For instance, if you want to use 4 cores, you would use the following command:

./mmp -t 4 genome.fasta reads.fastq

This produces a sam file printed on stdout. You can change the error rate with the option -e. For instance, if you know that it is 2%, then the command should be:

./mmp -e 0.02 genome.fasta reads.fastq

mmp is also compatible with gzipped input files (requires zcat in your $PATH):

./mmp -e 0.02 genome.fasta reads.fastq.gz