How to calculate P-value for biological repeats of ChIP-qPCR?
1
0
Entering edit mode
7.0 years ago
SP ▴ 280

This may be a very basic question but unfortunately I couldn't find an satisfactory answer. I want to calculate p-value for ChIP-qPCR %input between two conditions. How to normalize data for different biological repeats? Or should I just take the mean of %input?

chip-qpcr p-value • 3.8k views
ADD COMMENT
2
Entering edit mode

Time for a statistics lesson. a p-value is a probability of error from a hypothesis test. You've got to define what you mean to test and how for it to make any sense whatsoever.

ADD REPLY
0
Entering edit mode

I want to calculate p-value for difference in mean tag enrichment of a factor (DNA binding proteins) for chip-qpcr data at a genomic location in two conditions, A significant difference in mean tells me that binding of factor on genome has to do something with the biological condition. Let me know if it is clear enough.

ADD REPLY
1
Entering edit mode

That's better. a statistical test is about confirming the difference in a number between conditions. What makes up that number is not important. For you it seems to be "mean tag enrichment" however that's defined. You should get three measures of it in each b-replicate and try a Student's T-Test. That test assumes things come from normal/bell curve, so if that assumption is not reasonable, the test will be flawed. I don't know why you keep using the word mean though, seems to me "tag enrichment" at a location is just one number, mean implies you're collecting something?

ADD REPLY
0
Entering edit mode

You are correct it is tag enrichment but as you might know that every ChIP-qPCR biological replicate will have technical replicate, so I will take the mean of all technical replicate.

I was asking about the method for normalization before doing test like t-test because in ChIP-qPCR there are often a lot of variation between biological replicates. Normally in biological replicate you can see the pattern between wildtype sample with condition sample but due to high variation effects can not be find significant in statistical tests.

ADD REPLY
0
Entering edit mode
7.0 years ago

Use MA plots or better are volcano plots which also include the impression for significance.

Hint:

For volcano plots, calculate the t-test difference between replicates for control and sample conditions per locus, and then plot the t-test difference to the logP values from the t-test object itself (ex in R). Points on the top write and on top left (if you are asking, which locus protein loses its binding) are what you are looking for.

For more reading, consult:

ADD COMMENT

Login before adding your answer.

Traffic: 2524 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6