GQ

wendywendy taiwanMember

Hi,
I found some strange rules from my VCF file.
Most of snps which had high GQ, they also had high AF.
However, why the snp which had low AF,they got low GQ?
Does it make sense?

Thanks for your help.
Wendy

Best Answer

Answers

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin
    That does not seem surprising to me, but can you post a few examples of records showing this trend? If we have a concrete example we can discuss what may be going on in more detail.
  • wendywendy taiwanMember

    Hi Geraldine,

    This is the example of record showing this trend.
    We could find that the average of GQ increase as AF increase.
    However, in my opinion, when snps got low AF they also should get high GQ. Because that means it is easy to confirm the genotype.
    What is your opinion about this result?

    Thanks for your help.

    Wendy

    QD.jpg 63.1K
  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin
    I'm not sure I understand your reasoning. Assuming these are calls made from diploid data, it makes sense to me that whenever you see AF diverge strongly from the value that matches ideal ploidy, we will be less confident that we called the correct genotype, especially at low values. Have you read the docs that explain how these metrics are calculated?
  • wendywendy taiwanMember

    Hi Geraldine,

    It is true that these calls made from diploid data.

    However. I'm not sure what you mean about " it makes sense to me that whenever you see AF diverge strongly from the value that matches ideal ploidy, we will be less confident that we called the correct genotype, especially at low values."

    Could you possibly explain it in a little more detail?

    Thank you very much!

    Wendy

  • wendywendy taiwanMember

    Hi Sheila,
    These values were passed the filter
    --filterExpression "QD < 2.0 || FS > 60.0 || MQ < 40.0 || MQRankSum < -12.5 || ReadPosRankSum < -8.0"

    Also, I have another question. Why Most of snps which had high GQ, but they got low QD?
    Aren't we supposed that QD and GQ are positively correlated?

    Thank you very much!
    Wendy

    QD.jpg 35.5K
  • SheilaSheila Broad InstituteMember, Broadie, Moderator admin

    @wendy
    Hi Wendy,

    QD and GQ are different and not necessarily correlated. QD tells us how confident we are that there is actually a variant at the site. GQ tells us how confident we are in the genotype assigned to the sample. Have a look at this article.

    -Sheila

Sign In or Register to comment.