3
$\begingroup$

I have a qqplot of my whole genome sequencing data; A plot is for showing possibly significant driver genes. I tried to read about qqplot though but people only say about the skewedness while I want to know from these two genes which one are more likely to be driver

enter image description here

Improve this question
$\endgroup$

3 Answers 3

Reset to default
3
$\begingroup$

Given that TP53 is the most significant and is already known to have driver mutations in cancer it would seem to be the logical choice.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Thank you, does that mean I must ignore MAP1S as a candidate even though it has significant p-value but non-significant q-value? $\endgroup$
    – Angel
    Feb 18, 2019 at 14:57
  • 3
    $\begingroup$ Nothing I wrote suggests that you must ignore MAP1S, your question indicated you only wanted a single candidate. If you're fine with more then take more. $\endgroup$
    – Devon Ryan
    Feb 18, 2019 at 14:59
2
$\begingroup$

You can't know which is the driver and which is the carrier. At most you can say that a specific gene deviate more of the expected underlying hypothesis. See also other resources online.

You also seem to ignore other genes that deviate more of your null hypothesis. I recommend to plot the histogram of the p-values to see if the distribution is uniform

Improve this answer
$\endgroup$
3
  • $\begingroup$ I think it's pretty clear from the QQ-plot that these p-values are uniformly distributed (except for the twenty or so in the tail of the distribution)? $\endgroup$
    – winni2k
    Feb 20, 2019 at 15:34
  • $\begingroup$ @winni2k Usually the QQ-plot is centered around 0 (if using qqnorm), in this plot it is not centred, also a expected and observed pvalue above 1 indicates that this is not a normal QQ-plot. That's why I recommended the histogram. $\endgroup$
    – llrs
    Feb 20, 2019 at 16:01
  • $\begingroup$ Ah! I see. I think the x and y-axes must be -log10(p-value) as shown for example here, and not quantiles as would be reported by qqnorm. #missingaxislabels $\endgroup$
    – winni2k
    Feb 20, 2019 at 16:57
1
$\begingroup$

This looks to be a pretty good Q-Q plot. My recollection of doing Q-Q plots is that a good Q-Q requires a "more or less" linear relationship for the model to be considered okay. Again this model looks good, its 1:1, albeit it does deviate a little (for the gene of interest). TP15 falls outside the 1:1.

@Devon Ryan states TP15 has good biological credentials, which is clearly important. It maybe worth checking MAP15 because this is "more or less" within your model (for my Q-Q plots definitely within the model :-) ).

In summary, its just my opinion - that on the basis of the Q-Q plot the driver is the not the expected candidate.

Improve this answer
$\endgroup$

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