simulating illumina data with spiked variants

Question

[Richard]

Hi all,

I am looking to simulate some paired illumina data for a test. What I want to do in order of importance (most important at the top)

1. Create fastq files.
2. Specify specific SNPs to be in the data
3. Control the allelic fractions of the spiked in SNPs
4. Have an appropriate error model of illumina sequencing
5. Have controllable metrics like duplicate rate, chastity fail rate

There seem to be a number of tools available for simulating illumina - do we know of one that can handle my requirements?

Answer 1

[Devon Ryan]

Aside from #5, any of the common simulators (wgsim, Sherman, etc.) can do that. For the SNPs, just make a second genome containing them, sample from that as well, and then mix the results in the fraction that you'd like.

[Richard]

I checked both wgsim and sherman and I didn't see a way to spike in specific variants (base change and position). Am I missing something?

[Devon Ryan]

Read the entirety of my answer, I mentioned the variants explicitly.

Answer 2

[Gabriel R.]

For 1, 2 (maybe 3), 4 we developed a sequencing simulator for ancient DNA: grenaud.github.io/gargammel/ But it can be used to simulate modern DNA. For the PhiX, you can add it as a "microbial contaminant". It automates the process of fragment size distribution, sampling from a diploid genome and generate Illumina-like fastq files. Just give it a "diploid" genome represented by 2 fasta files.