ADRSM (Ancient DNA Read Simulator for Metagenomics) is a tool designed to simulate the paired-end sequencing of a metagenomic community. ADRSM allows you to control precisely the amount of DNA from each organism in the community, which can be used, for example, to benchmark different metagenomics methods.

• Conda

conda install -c maxibor adrsm

adrsm ./data/short_genome_list.csv

• metagenome.{1,2}.fastq : Simulated paired end reads
• stats.csv : Statistics of simulated metagenome (organism, percentage of organism's DNA in metagenome)

You can cite ADRSM like this:

Maxime Borry (2018). ADRSM: Ancient DNA Read Simulator for Metagenomics. DOI: 10.5281/zenodo.1462743

$ adrsm --help
usage: ADRSM v0.9 [-h] [-r READLENGTH] [-n NBINOM] [-fwd FWDADAPT]
              [-rev REVADAPT] [-e ERROR] [-p GEOM_P] [-m MIN] [-M MAX]
              [-o OUTPUT] [-s STATS] [-se SEED] [-t THREADS]
              confFile

==================================================

ADRSM: Ancient DNA Read Simulator for Metagenomics

Author: Maxime Borry

Contact: <borry[at]shh.mpg.de>

Homepage & Documentation: github.com/maxibor/adrsm
==================================================


positional arguments:
confFile       path to configuration file

optional arguments:
-h, --help     show this help message and exit
-r READLENGTH  Average read length. Default = 76
-n NBINOM      n parameter for Negative Binomial insert length distribution.
                Default = 8
-fwd FWDADAPT  Forward adaptor. Default = AGATCGGAAGAGCACACGTCTGAACTCCAGTCAC
                NNNNNNATCTCGTATGCCGTCTTCTGCTTG
-rev REVADAPT  Reverse adaptor. Default =
                AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGTAGATCTCGGTGGTCGCCGTATCATT
-e ERROR       Illumina sequecing error. Default = 0.01
-p GEOM_P      Geometric distribution parameter for deamination. Default =
                0.5
-m MIN         Deamination substitution base frequency. Default = 0.001
-M MAX         Deamination substitution max frequency. Default = 0.3
-o OUTPUT      Output file basename. Default = ./metagenome.*
-s STATS       Statistic file. Default = stats.csv
-se SEED       Seed for random generator. Default = 7357
-t THREADS     Number of threads for parallel processing. Default = 2

The configuration file is a .csv file describing, one line per genome, the mean insert size, and the expected genome coverage. Example short_genome_list.csv:

Given the sequencing error, and the random choice of inserts, the target coverage might differ slightly from the real coverage (fig 1)

Figure 1: Coverage plot for simulated sequencing of Elephas maximus mitocondria. Aligned with Bowtie2 (default-parameters). Read-length = 76, insert-length = 200.

The deamination is modeled using a Geometric distribution With the default parameters, the substitution frequency is depicted in fig 2:

One can try different parameters for deamination using this interactive plot: maxibor.github.io/adrsm

Figure 2: Substitution frequency.

For each nucleotide, a random number Pu is sampled from an uniform distribution (of support [0 ,1]) and compared to the corresponding value Pg of the rescaled geometric PMF at this nucleotide.
If Pg >= Pu, the base is substituted (fig 3).

Figure 3: Substitutions distribution along a DNA insert, with default parameters.

ADRSM can simulate Illumina sequencing error with a uniform based model.

The base quality score is generated using a Markov chain from fastq template files.

ADRSM offers you to add mutation to your sequences. This allows to account for the evolutionary differences between ancient organisms and their reference genome counterparts present in today's databases.

ADRSM assumes two times more transitions than transversions.

There are two parameters for mutation simulation:

• The mutation rate (in bp/year): a good starting point is 10e-7 for bacteria
• The age (in years) of the organism