pbsim(1) simulator for PacBio sequencing reads


pbsim [options] <reference.fasta>


The pbsim command produces simulated PacBio reads for reference FASTA sequence <reference.fasta>.

Model files (parameters for the --model-qc option) can be found in the /usr/share/pbsim/models directory.


The options for pbsim can be divided into general, sampling-based and model-based simulation options.

General options


prefix of output files (sd).


data type. CLR or CCS (CLR).


depth of coverage (CLR: 20.0, CCS: 50.0).


minimum length (100).


maximum length (CLR: 25000, CCS: 2500).


minimum accuracy (CLR: 0.75, CCS: fixed as 0.75). This option can be used only in case of CLR.


maximum accuracy (CLR: 1.00, CCS: fixed as 1.00). This option can be used only in case of CLR.


ratio of differences. substitution:insertion:deletion. Each value up to 1000 (CLR: 10:60:30, CCS:6:21:73).


for a pseudorandom number generator (Unix time).

Options for sampling-based simulation


FASTQ format file to sample.


sample-fastq (filtered) profile ID. When using --sample-fastq, profile is stored. sample_profile_<ID>.fastq, and sample_profile_<ID>_.stats are created. When not using --sample-fastq, profile is re-used. Note that when profile is used, --length-min,max, --accuracy-min,max would be the same as the profile.

Options for model-based simulation


model of quality code.


mean of length model (CLR: 3000.0, CCS:450.0).


standard deviation of length model (CLR: 2300.0, CCS: 170.0).


mean of accuracy model (CLR: 0.78, CCS: fixed as 0.98). This option can be used only in case of CLR.


standard deviation of accuracy model (CLR: 0.02, CCS: fixed as 0.02). This option can be used only in case of CLR.


To run model-based simulation:

pbsim --data-type CLR \
      --depth 20 \
      --model_qc /usr/share/pbsim/models/model_qc_clr \

In the example above, simulated read sequences are randomly sampled from a reference sequence ("reference.fasta") and differences (errors) of the sampled reads are introduced. Data type is CLR, and coverage depth is 20. If the reference sequence is multi-FASTA file, the simulated data is created for each FASTA. Three output files are created for each FASTA. "sd_0001.ref" is a single-FASTA file which is copied from the reference sequence. "sd_0001.fastq" is a simulated read dataset in the FASTQ format. "sd_0001.maf" is a list of alignments between reference sequence and simulated reads in the MAF format. The length and accuracy of reads are simulated based on our model of PacBio read.

To run sampling-based simulation:

pbsim --data-type CLR \
      --depth 20 \
      --sample-fastq sample.fastq \

In the sampling-based simulation, read length and quality score are the same as those of a read taken randomly in the sample PacBio dataset ("sample.fastq").


pbsim is available under the terms of the GNU General Public License, version 2 (GPL-2).


Michiaki Hamada ( Yukiteru Ono