PRESUME is a software tool that simulates cell division or speciation and diversification of DNA sequences in the growing population. By employing a distributed computing platform, PRESUME progressively generates a large number of sequences that accumulate substitutions with their lineage history information. In this process, daughter sequences are duplicated at a certain speed which is incompletely inherited from that of the maternal sequence under a stochastic model (Figure 1). The substitution probability at different positions in each sequence is defined in a time-dependent manner using GTR-Gamma model or set to a certain rate. The software allows the user to simulate various types of sequence diversification processes with different sets of input parameters.
Figure 1. Schematic diagram of PRESUME. PRESUME simulates the propagation and diversification of sequences that accumulate substitutions and generates a large set of descendant sequences with lineage information. m refers to a maternal sequence, and d1 and d2 refers to two daughter sequences derived from m. In this simulation, the doubling times of the two daughter sequences (td1 and td2) are incompletely inherited from the doubling time of the mother sequence (tm). This occurs under a stochastic model, in which 1/td1 and 1/td2 follow a normal distribution where the mean and variance are 1/tm and σ2 respectively. Additionally, sequence extinction is set at a random rate (ε) and also occurs when the sequence doubling speed reaches a negative value. The substitution probabilities at different positions in each sequence of length L are defined in a time-dependent manner using GTR-Gamma model with parameters Q, α and μ, or set to a certain rate φ (see SubstitutionModelDetails.PRESUME.pdf).
- PRESUME can be executed on MacOS or Linux.
- The distributed computing mode of PRESUME requires UGE (Univa Grid Engine)
- Python3 (newer than 3.7.0) with Biopython module required and tqdm module optional; if you want to visualize a simulation progress
Each step of installation takes less than 1 min.
-
Download PRESUME by
git clone https://github.com/yachielab/PRESUME -
Add the absolute path of PRESUME directory to $PATH
-
Make PRESUME executable
chmod u+x PRESUME.py
Installation of Anaconda (required)
-
Execute the following commands
wget https://repo.anaconda.com/archive/Anaconda3-2018.12-Linux-x86_64.sh bash Anaconda3-2018.12-Linux-x86_64.sh -
Set $PATH to anaconda3/bin
Installation of Biopython 1.76 (required)
-
Install Biopython by
conda install -c anaconda biopython
Installation of tqdm 4.43.0 (optional)
-
Install tqdm by
conda install -c conda-forge tqdm
The software functions can be tested by the following example commands:
Example 1
Generation of ~100 sequences using GTR-Gamma model with the default parameter set without distributed computing. The computation will take several minutes.
PRESUME.py -n 100 --gtrgamma default --save
Output: a directory PRESUMEout containing the following files will be created in your working directory:
-
PRESUMEout.fa: FASTA file for generated descendant sequences -
root.fa: FASTA file describing the root sequence used for the simulation -
PRESUMEout.nwk: Newick format file for the lineage history of the generated sequences -
args.csv: CSV file containing basic patameters used for the simulation (enabled by --save).
Example 2
Generation of ~100 sequences using a time-independent model with the substitution frequency of 5% per site per generation along with a highly unbalanced lineage trajectory (σ of 10). The computation will take several minutes.
PRESUME.py -n 100 --constant 0.05 -s 10
Output data: a directory PRESUMEout will be created in your working directory.
Example 3
Generation of ~10,000 sequences using GTR-Gamma model with a defined parameter set with distributed computing. The computation will take several minutes.
PRESUME.py -n 10000 --gtrgamma GTR{0.927000/2.219783/1.575175/0.861651/4.748809/1.000000}+FU{0.298/0.215/0.304/0.183}+G{0.553549} --qsub
Output data: a directory PRESUMEout will be created in your working directory.
Example 4
Generation of ~100 sequences using GTR-Gamma model and an original indel model with a defined parameter set with distributed computing. The computation will take ~1 minute.
PRESUME.py -n 100 --gtrgamma default --inprob prob.txt --inlength length.txt --delprob prob.txt --dellength length.txtInput data: prob.txt defines indel probability per generation for each initial sequence postion and length.txt defines the distribution of each indel
Output data: a directory PRESUMEout will be created in your working directory.
See SubstitutionModelDetails.PRESUME.pdf for more details of how to specify the GTR-Gamma model parameters.
Note that as the number of sequences are only sporadically monitored during the simulation, the number of generated descendant sequences can be fluctuated and differed from the number of sequences N required to be generated by -n.
In the distributed computing mode, the number of jobs will be around √N; PRESUME first generates ~√N number of sequences in a single node, each of which is then subjected to the further downstream process in a distributed computing node.
Usage:
PRESUME.py
[-v] [--version] [-h] [--help] [-n sequence_number] [-L sequence_length] [-s standard_deviation]
[-e extinction_probability] [--gtrgamma model_parameters] [-m mean_substitution_rate]
[--constant substitution_probability] [--qsub] [--output directory_path] [-f input_file]
[--load file_name] [-u sequences_number] [--debug] [--bar] [--save]
[-r max_retrial_number] [--seed random_seed] [--limit time_limit]
Options:
-v --version
Print PRESUME version; ignore all of the other parameters
-h --help
Print the usage of PRESUME; ignore all of the other parameters
-n <Integer>
Number of sequences to be generated. Default: 100
-L <Integer>
Length of sequences to be generated. Default: 1000
-s <Float>
Standard deviation of propagation speed. Default: 0
-e <Float>
Probability of extinction. Default: 0
--gtrgamma <String>
GTR-Gamma model parameters
Format： --gtrgamma GTR{A-C/A-G/A-T/C-G/C-T/G-T}+FU{piA/piC/piG/piT}+G{alpha}
For more details, see https://github.com/yachielab/PRESUME/blob/master/SubstitutionModelDetails.PRESUME.pdf
or you can use the default parameter set by
--gtrgamma default
which is equivalent to
--gtrgamma GTR{0.03333/0.03333/0.03333/0.03333/0.03333/0.03333}+FU{0.25/0.25/0.25/0.25}+G4{10000}
-m <Float>
Mean of gamma distribution for relative substitution rates of different sequence
Positions. Default: 1
--constant <Float>
Execute time-independent model to simulate sequence diversification with a parameter of
constant substitution probability per generation of every sequence position
--qsub
Execute the distributed computing mode
--output <String>
Output directory path. PRESUME creates a directory unless exists. Default: current directory
-f <String>
Input FASTA file name for the root sequence. Random sequence will be generated unless specified
-u <Integer>
Maximum number of sequences to be generated. Default: 1000000000
--debug
Output intermediate files
--bar
Activate the monitoring of simulation progress with Python tqdm module
--save
Output a CSV file for parameter values used for the simulation
--param <String>
CSV file for parameter values.
This file can be obtained from a previous simulation run executed with –-save option.
-r <Integer>
Maximum number of retrials of simulation when all sequences are extinct
--seed <Integer>
Seed value for generation of random values. Default: 0
--monitor <float>
Stepper size parameter for monitoring of lineage generation. Default: 1
--tree <String>
Input Newick format file name if a template tree is given.
The following parameters will be ignored:
-L -s -e -f -u -r --constant –-qsub –-load –-debug –-bar –-save –-seed --limit
- Keito Watano (The University of Tokyo) watano.k10.yachielab@gmail.com
- Naoki Konno (The University of Tokyo) naoki@bs.s.u-tokyo.ac.jp
- Nozomu Yachie (The University of Tokyo) nzmyachie@gmail.com