Simulate short-read RNA-seq data from long-read RNA-seq data

Question

[rhonddaskl]

Hi,

I'm trying to use my self-developed tool to test it on long-read and short-read single-cell RNA-seq datasets. To ensure a fair comparison, I'm attempting to simulate short-length single-cell RNA-seq data from the long-read RNA-seq data. For example, for each sequence in the long-read FASTQ file, I will randomly select sequences that are 100 bp long to simulate a Smart-seq dataset. I will then use my tool for comparison. I'm wondering if there is a tool that can perform a similar task. I've noticed some people using Polyester to achieve this with transcript datasets, but I'm curious if I can directly use the FASTQ data to simulate short-read RNA-seq.

thank you!

Best, R

Answer 1

[Mensur Dlakic]

I think you could try doing this, but it wouldn't be appropriate. You'd be bringing long-read errors, which I believe are higher than short-read errors, into your "simulated" short reads. Separately, you wouldn't be creating any type of error that are hallmarks of short reads, so this would be a short-read sample only in name.

If I were to do this I'd assemble long reads first which will correct some of those errors, and simulate from the assembled material.

[rhonddaskl]

thank you for your suggestion!