No announcement yet.

What is the meaning of quality value?

  • Filter
  • Time
  • Show
Clear All
new posts

  • What is the meaning of quality value?

    In the XXXX_.QV.qual file, there is a quality value for each position.

    I wonder how they calculate this value and what is the meaning of this value.

    For traditional sequecing, I think quality value is the P value of the corret read over the noise, then -10logP. However, for SOLID, every read is from 2 base pair, how do they get the quality value? Anybody can provide some advice or reference?

    Thanks a lot!

  • #2
    I think it gives quality value for color call...this is abased many parameters majorly intensity.


    • #3
      This is all I have been able to find:

      "The QV value is calculated using a phred like score
      q = -10 x log10(p)
      where q is the quality value and (p) is the predicted probability that the color call is incorrect"

      which is from this document:


      • #4
        If you convert from colorspace to [ATGC], how do you convert the QV value? Take the average?


        • #5
          Originally posted by juan View Post
          If you convert from colorspace to [ATGC], how do you convert the QV value? Take the average?
          No, that would not be right.
          I recommend against conversion of raw reads from color space because an error in one position would propagate through the rest of the read. That is, if your read had a single error at base 2, then by converting the read to base space, you would ensure that every base after base 2 was incorrect.

          Later errors could shift you back into the correct sequence frame, however. So, even were you wanting to risk switching into base space, there would be no way to convert the quality values. Or at least, no trivial way.

          On the other hand, converting base space sequence into color space involves no such risk. Converting quality values would still not be straight forward, but you could estimate them by choosing the lower quality value of the two bases.

          Really, if you are trying to use color space-naive code, your only decent choice is to use "double encoding". This involves doing a tr/0123/acgt/ conversion. The resulting fake sequence now is usable in color space-naive code. With the caveat that reverse-complementing is illegitimate on double encoded sequence. (Because in color space one reverses to reverse-complement...)