Illumina GAIIx sequecing error rate (background)

[afalvarez]

Hi guys,

We are using NGS to detect non-clonal, low frequency mutations, so we are quite concerned about the background the sequencer introduces, meaning the amount of inespecific mutations coming from the sequencing process rather than from our samples.

In that sense, all the information we've found refers to the error rates coming from base calling (99.9% accuracy for GAIIx, 99.99% for SOLiD, etc.), but nothing about the errors generated during the sequencing process per se (library preparation, cluster generation, SBS...).

Do you know if there is any experimental data available? For instance, a measure of the mutational load of a known sequence (let's say the phage PhiX genome -usually loaded as a calibration control-) in subsequent technical replicates, or something like that.

Thanks a lot!

[nucacidhunter]

Here is a link for an earlier publication: http://genomebiology.com/2011/12/11/R112. There might be more information in papers citing this one.

[Brian Bushnell]

...all the information we've found refers to the error rates coming from base calling (99.9% accuracy for GAIIx, 99.99% for SOLiD, etc.)

I think you should be very wary of that particular source of information

[afalvarez]

Manufacturer's enthusiasm

I think you should be very wary of that particular source of information

For sure. Anyway, look at what AB claims in the SOLiD 5500 whitepaper:

"As the results demonstrate, Exact Call Chemistry determines the
template sequence with extremely high accuracy, the majority
of base calls achieving accuracy in excess of 99.9999%"

And 99,99% came from the specification sheet of the very same sequencer...

Manufacturers are usually too enthusiastic about their products

[afalvarez]

Here is a link for an earlier publication: http://genomebiology.com/2011/12/11/R112. There might be more information in papers citing this one.

Thanks!

As a quick reference to the data that can be found in the paper, these are the substitution rates found in PhiX genome:

1) HiSeq

a. 2x95 (4.3e6 pair reads; spiked) --> 0.11% substitution errors
b. 2x100 (8.87e5 pair reads; spiked) --> 0.12% substitution errors

2) GAIIx

2x150 (6.4e6 pair reads; one lane) --> 0.28% substitution errors

These numbers are worked out as (Total errors) / (Uniquely aligned bases) using B-tail trimmed reads.

Error rates with different quality filters can be found in Table6, it worths having a look at them as well.