How to detect random mutations by genome resequencing?

[forestsong]

Now being puzzled by a (maybe stupid) question. Any comments or suggestions would be appreciated.

A certain drug treatment on my cultured cells induces highly upregulated expresion of the APOBEC deaminase protein family. I would like to know whether this upregulation will lead to enhanced deamination targeting the cells' genome.

Since the deamination by APOBEC deaminase (inducing a C to T mutation) is expected in a random manner distributed within the whole genome, and with a relatively low frequency, so how to distinguish a real deamination from the common sequencing error (>0.1%, could be even much higher than the deamination mutation rate). I just feel it somehow a mission impossible.

I hope I explained the question well and thanks a lot for your help in advance!

[krobison]

A few thoughts:

Since you are trying to calculate overall mutation rates, you might want to scan carefully a small amount of the genome rather than actually trying to scan everything. So, two options are RAD-Seq and amplicon sequencing.

Using paired-end Illumina sequencing with relatively short inserts (about your read length) would enable higher accuracy by reading each base of a fragment with both reads.

There have been a number of papers which have used embedded random tags in either primers or adapters to drive for higher accuracy. The idea is to deliberately oversample the diversity of the library and call mutations only if the same fragment (as identified by the random tags) is sequenced multiple times and the mutation is called consistently.

[forestsong]

Thank you very much, Krobison!

I also thought about scanning a part of the genome in deep (eg. exon sequencing) in stead of the whole picture, but I do worry about the criticism that those narrowed pieces will not be representative / convincing enough to exclude the possible low-rate random mutations.

On the other hand, considering the mutations induced by APOBEC deamination would be totally individual, and probably spread without any "hotspot" in the genome, it can hardly find a second "call" by the coverage I could reach based on my budget. Maybe I didn't get your point about the idea "embedded random tags". Could you please kindly provide a reference?

Thank you again for your kind help!

[krobison]

yes, if you focus there is a potential criticism that these are biased; one potential argument for RAD-Seq is that these aren't strictly human-picked, so you can't be accused of cherry-picking.

Some references for the tagging approach

Detection and quantification of rare mutations with massively parallel sequencing.

Accurate sampling and deep sequencing of the HIV-1 protease gene using a Primer ID.

Digital RNA sequencing minimizes sequence-dependent bias and amplification noise with optimized single-molecule barcodes.

A method for counting PCR template molecules with application to next-generation sequencing

Detection of ultra-rare mutations by next-generation sequencing.

[forestsong]

Those are quite helpful. Thanka a lot!