SimLoRD

Simulate PacBio SMRT long reads (with pass-dependent insertion, deletion and substitution errors) from a reference FASTA.

License: MIT. Source code.

What it simulates

SimLoRD simulates PacBio SMRT long reads from a reference FASTA. Its error model follows the number of passes a molecule gets around the circular template. More passes give a more accurate consensus read. Each read draws a pass count, and the insertion, deletion and substitution probabilities fall with it. You give a reference, a coverage and a read-length rule. It returns a FASTQ, and optionally a SAM of the true alignments.

Key assumptions: the single-pass error rates are the ones you set (defaults 0.11 insertion, 0.04 deletion, 0.01 substitution), and the pass count follows a fitted chi-squared model capped by --max-passes (default 40).

When to use it

Test a PacBio-oriented pipeline across a range of accuracy, from noisy single-pass to near-perfect consensus reads.
Draw read lengths from a log-normal distribution, a fixed length, or the lengths in your own FASTQ.
Get the true origin of each read from the header.

Alternatives on biosimhub: PBSIM2 and PBSIM3 use quality-score HMMs and cover Nanopore too. CycSim learns errors from real reads and ties them to the local sequence.

Install it yourself

SimLoRD 1.0.4 is on bioconda (dry-run resolved on linux-aarch64 to build py310h5cfe7d9_6, 34 kB).

conda create -n simlord -c conda-forge -c bioconda simlord=1.0.4
conda activate simlord

Minimal command:

simlord --read-reference reference.fa --coverage 5 --fixed-readlength 1000 --no-sam sim

The last argument is the output prefix. The reads go to sim.fastq. Without --no-sam it also aligns every read and writes sim.sam. Other options: --num-reads (default 1000) instead of --coverage, --gzip, --without-ns, and --generate-reference GC LENGTH to make a random genome.

Run the tested example

This recipe builds SimLoRD in a container and runs the example below. It needs Docker and nothing else. It last ran on 2026-10-12 in 0.8 s.

Download simlord-recipe.tar.gz (Dockerfile, example reference.fa, run.sh).

Unpack it and run the script:

tar -xzf simlord-recipe.tar.gz
cd simlord
sh run.sh

What run.sh does

docker build -t biosimhub/simlord:1.0.4 .
mkdir -p out && chmod a+w out
docker run --rm \
  --user "$(id -u):$(id -g)" -e HOME=/tmp \
  -v "$PWD/in:/work/in:ro" -v "$PWD/out:/work/out" -w /work/out \
  biosimhub/simlord:1.0.4 simlord --read-reference /work/in/reference.fa \
    --coverage 5 --fixed-readlength 1000 --prob-ins 0.11 \
    --prob-del 0.04 --prob-sub 0.01 --max-passes 40 \
    --no-sam sim > out/simlord_log.txt

Inputs are read from in/ and results land in out/. Change the values after each option to run your own case; the parameter table lists them.

Expected output

File	Size
`sim.fastq`	213.6 KB
`simlord_log.txt`	0 B

First lines of sim.fastq

@m0/1000/CCS Read=0;length=1000bp;startpos=2421;chromosome=synthetic_ref;numberOfErrors=21;totalErrorProb=0.0271;passes=3.672087738420167;passesLeft=3;passesRight=4;cutPosition=328
AGTGTTCGCATCTGGCGCTCAGATTTTAAGTTAATCTGCCGAACCGTCGGCTTTCGCAGGTCCAGTATCACTGTCTTACAGGAGTGCGACTAGGGCCCGCGCAGTGTGAACGATTGTATGGGGCGAGTGATGTTAGTGGAAAGGCAGCTCATTCTAGCCGACGTGGCCAATGCCTAACCACTCCGTGCCGTAGATACGTA
+
22222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222
@m1/1000/CCS Read=1;length=1000bp;startpos=4665;chromosome=synthetic_ref;numberOfErrors=2;totalErrorProb=0.0010;passes=9.453415682616459;passesLeft=9;passesRight=10;cutPosition=547
TTAGTTAAGAGGGTGCAAGCGGAGTGAAAATCGG

Parameters

Option	Meaning	Example value	Range
`--coverage`	Coverage	5	0.1 to 50
`--fixed-readlength`	Fixed read length	1000	100 to 20000
`--prob-ins`	Insertion probability (single pass)	0.11	0 to 0.3
`--prob-del`	Deletion probability (single pass)	0.04	0 to 0.3
`--prob-sub`	Substitution probability (single pass)	0.01	0 to 0.3
`--max-passes`	Maximum number of passes	40	1 to 100

Ranges are the limits used for the example recipe; the program itself may accept more.

Reading the output

The reference is one 20,000 bp sequence named synthetic_ref. At 5x with fixed 1000 bp reads, SimLoRD writes 100 reads (100,000 bases) into sim.fastq (218,756 bytes). simlord_log.txt is empty.

The first header reads:

@m0/1000/CCS Read=0;length=1000bp;startpos=2421;chromosome=synthetic_ref;numberOfErrors=21;totalErrorProb=0.0271;passes=3.67;passesLeft=3;passesRight=4;cutPosition=328

startpos is the 0-based origin on the reference.
numberOfErrors=21 is the count of inserted, deleted and changed bases in this read.
passes=3.67 is the pass count. Fewer passes mean more errors. The second read has 9.45 passes and 2 errors.
Quality values are one flat character (2, Phred 17) on every base. They carry no per-base information.

Across the 100 reads the mean is 4.98 errors per 1000 bp and 9.88 passes.

Your own runs

Single pass. Add --max-passes 1. Result: 100 reads, all with passes=1, and a mean of 173.0 errors per read (17.3%) against 4.98 before.
Real-looking lengths. Replace --fixed-readlength 1000 with --lognorm-readlength and use coverage 10. Result: 24 reads, 212,073 bases, mean 8,836 bp. Coverage targets total bases (10 x 20,000 = 200,000), so a long-tailed length draw gives fewer reads.
Your genome. Mount your FASTA, keep --no-sam while you tune, and drop it when you need truth alignments.

Pitfalls

Error rates are single-pass rates. --prob-ins, --prob-del and --prob-sub describe a one-pass read. Multi-pass reads get fewer errors, so the reads you see are cleaner than those values.
Flat quality scores. Tools that use base qualities see no signal here. Do not use these FASTQs to test quality filtering.
Coverage is by bases. The read count depends on the length rule. Use --num-reads for an exact count.
SAM output costs time. Without --no-sam, SimLoRD aligns every read and writes a SAM file. Large genomes need the extra time and disk.
No seed option. The help lists no seed flag, so runs are not repeatable.