simulate illumina WGS run

[nheb]

Hello everyone

I am wondering if anyone could suggest any scripts to do the following:

simulate NGS data from a particular genome with a given coverage.

For example, given mouse genome (or any other sequence), create WGS reads with X depth from illumina. (IT would be great if it mimics the error distribution).

[nheb]

in-silico BAC library

I am looking to generate a simulation of BAC library with a given BAC size distribution with a certain X coverage.

For example, cut a genome into sequences of certain length range to get a required coverage.

Thanks a lot.

[Brian Bushnell]

I have a nice tool for this purpose in the BBTools package. An example command:

randomreads.sh ref=reference.fa out=reads.fq reads=1000000 length=100 paired mininsert=100 maxinsert=300 gaussian minq=20 midq=30 maxq=36

The reads will come out annotated with their genomic origin. You can add snps and indels if you want, as well. The error profile is pretty close to the real Illumina error model. Run randomreads.sh with no arguments for more information.

[GenoMax]

ART is another option for read simulation.

[luc]

ART is more complex and ambitious. I could not get it to run on for bigger genomes (1.5Gb) though. Did anybody else have better luck?

[Brian Bushnell]

ART is more complex and ambitious.

Are you sure about that?

[luc]

Are you sure about that?

Hi Brian,

I hope you will show me otherwise.

Among other ART options there are insert size variation (stdev) , various error profiles separate for forward and reverse reads, sam file output.
However it is slow and did not work at all for me for a 1.5 Gb genome.


From the ART readme:

===== PARAMETERS =====

-1 --qprof1 the first-read quality profile
-2 --qprof2 the second-read quality profile
-amp --amplicon amplicon sequencing simulation
-d --id the prefix identification tag for read ID
-ef --errfree indicate to generate the zero sequencing errors SAM file as well the regular one
NOTE: the reads in the zero-error SAM file have the same alignment positions
as those in the regular SAM file, but have no sequencing errors
-f --fcov the fold of read coverage to be simulated or number of reads/read pairs generated for each amplicon
-h --help print out usage information
-i --in the filename of input DNA/RNA reference
-ir --insRate the first-read insertion rate (default: 0.00009)
-ir2 --insRate2 the second-read insertion rate (default: 0.00015)
-dr --delRate the first-read deletion rate (default: 0.00011)
-dr2 --delRate2 the second-read deletion rate (default: 0.00023)
-l --len the length of reads to be simulated
-m --mflen the mean size of DNA/RNA fragments for paired-end simulations
-mp --matepair indicate a mate-pair read simulation
-nf --maskN the cutoff frequency of 'N' in a window size of the read length for masking genomic regions
NOTE: default: '-nf 1' to mask all regions with 'N'. Use '-nf 0' to turn off masking
-na --noALN do not output ALN alignment file
-o --out the prefix of output filename
-p --paired indicate a paired-end read simulation or to generate reads from both ends of amplicons
-q --quiet turn off end of run summary
-qs --qShift the amount to shift every first-read quality score by
-qs2 --qShift2 the amount to shift every second-read quality score by
NOTE: For -qs/-qs2 option, a positive number will shift up quality scores (the max is 93)
that reduce substitution sequencing errors and a negative number will shift down
quality scores that increase sequencing errors. If shifting scores by x, the error
rate will be 1/(10^(x/10)) of the default profile.
-rs --rndSeed the seed for random number generator (default: system time in second)
NOTE: using a fixed seed to generate two identical datasets from different runs
-s --sdev the standard deviation of DNA/RNA fragment size for paired-end simulations.
-sam --samout indicate to generate SAM alignment file
-sp --sepProf indicate to use separate quality profiles for different bases (ATGC)
NOTE: art will automatically switch to a mate-pair simulation if the given mean fragment size >= 2000