7
$\begingroup$

I'd like to simulate 10% sequencing error using art_illumina. The simulator doesn't have a parameter that I can just give the 10%, but it has this:

-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.

Heng Li's wgsim has an option for base error rate. Can I do the same for art_illumina?

Q: Does the "error rate" in the description mean sequencing error? If I set both qShift and qShift2 to 10, 1/(10^(x/10)) == 0.1, where x is 10, does that mean 10 is the value I should give to the simulator?

$\endgroup$
1
  • $\begingroup$ "a positive number will shift up quality scores that reduce substitution sequencing errors" -- I suspect you want to do the opposite: "a negative number will shift down quality scores that increase sequencing errors" $\endgroup$
    – gringer
    Jul 5 '17 at 8:44
4
$\begingroup$

You need to specify an error profile to do this. However, you can shift quality scores of an existing profiles to get 10% error, which is easier than defining a complete profile.

For example, Illumina HiSeq 1000 has the error rate of 0.1% and you want 100 time more errors, so you should shift down the quality scores by 20 by setting both qShift and qShift2 to -20. The following code will simulate paired end reads with 10% error.

art_illumina -ss HS10 -i reference.fa -qs -20 -qs2 -20 -sam -p -l 100 -f 20 -m 200 -s 10 -o output
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.