Question: What statistical test should I apply to prove my high confidence values are part of low confidence values?
0
gravatar for morovatunc
2.9 years ago by
morovatunc380
Turkey
morovatunc380 wrote:

Dear all,

I got chip seq peak results with bed format. There are 7 normal vs 13 tumor samples. I have two consensus peak results from this data.

1) I calculated overlap regions of normal and tumor samples separately first. After that, I compared two groups. This is my high confidence peaks.

2)I piped merged all normal files and tumor files into separate bed then, calculated the overlaps. This is my low confidence peaks.

I wanted to show that 1st one is part of my 2nd group statistically. I decided to apply t test but I couldn't figure out a way to implement it to bed files. Could you show me way to do that?

 

Thank you for your help.

Best regards,

Tunc

 

 

 

 

 

 

chip seq bed format • 1.1k views
ADD COMMENTlink modified 2.9 years ago • written 2.9 years ago by morovatunc380
1
gravatar for yupeng he
2.9 years ago by
yupeng he80
United States/San Diego/UCSD
yupeng he80 wrote:

It does not look like that a statistical test is needed here. If I understand correctly, all of your high confidence peak will be in the low confidence peak. There is no randomness here. 

Why do you want to test the overlap? Is there any randomness (e.g. random error or variation due to sampling) you are worrying about?

ADD COMMENTlink written 2.9 years ago by yupeng he80

actually there is not any randomness. I just want to explain the relationship with low vs high confidences. For example, some percent of my low confidence values is also shown in high confidence etc.. My question might be very stupid therefore I apologise if I ask something too basic or stupid. 

 

ADD REPLYlink written 2.9 years ago by morovatunc380

No worries. I was trying to understand. Reporting the percentages should be ok. For example, the fraction of low confidence peaks is high confidence peaks. 

ADD REPLYlink written 2.9 years ago by yupeng he80

If there is no randomness, a test is not needed. Description should be good enough. If you are curious about the case with randomness, you may want to look into proportion test:

http://stattrek.com/hypothesis-test/proportion.aspx?Tutorial=AP

ADD REPLYlink written 2.9 years ago by yupeng he80
0
gravatar for andrew.j.skelton73
2.9 years ago by
London
andrew.j.skelton735.3k wrote:

Hi, did you see this post a couple of days ago? May be useful for your problem!

ADD COMMENTlink written 2.9 years ago by andrew.j.skelton735.3k
0
gravatar for morovatunc
2.9 years ago by
morovatunc380
Turkey
morovatunc380 wrote:

Thank you for your help ! I will read this topic and learn more about the concepts.

ADD COMMENTlink written 2.9 years ago by morovatunc380
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1400 users visited in the last hour