## User: yjiangnan

yjiangnan •

**10**- Reputation:
**10**- Status:
- New User
- Location:
- Last seen:
- 12 months ago
- Joined:
- 1 year, 2 months ago
- Email:
- y********@gmail.com

#### Posts by yjiangnan

<prev
• 20 results •
page 1 of 2 •
next >

0

votes

3

answers

2.2k

views

3

answers

... Hi Robert, Thank you very much for the additional plots. Moderated t-test in your plots seems to have worked well at the 5% FDR level, but it also increased -log10 p-values for non-sex-specific genes so that several non-sex-specific genes were "significant" at the 10% FDR level. If you went further ...

written 14 months ago by
yjiangnan •

**10**0

votes

3

answers

2.2k

views

3

answers

... Then you confused me. If DESeq is more conservative, how did you get the much larger p-value? Did you only count their data as two replicates (with duplicates combined) each age group, or did you treat their data as 4 replicates each age group? Or, do you think they just incorrectly used the DESeq p ...

written 14 months ago by
yjiangnan •

**10**0

votes

3

answers

2.2k

views

3

answers

... Yes, that is the gene I was talking about. Your result is interesting. It confirms that the original paper had problems with at least this gene and likely with other genes because the p-values were so different.
However, DESeq2 seems just a newer version of DESeq to any non-specialist. Even in your ...

written 14 months ago by
yjiangnan •

**10**0

votes

3

answers

2.2k

views

3

answers

... I specified; you missed it. All samples were assumed to be drawn from the same population, that necessarily means that we were talking about testing the sample mean against the population mean by default by any person that has a basic understanding in statistics. Null hypothesis is that the sample m ...

written 14 months ago by
yjiangnan •

**10**0

votes

3

answers

2.2k

views

3

answers

... I disagree. If you deviate from this basic property of p-values, your statistics would dangerously go to the path of being useless in science because it no longer reflects reproducibility. Maybe I have to further clarify what I meant by "converge", or it may not be the best word. If you test 1 milli ...

written 14 months ago by
yjiangnan •

**10**0

votes

3

answers

2.2k

views

3

answers

... Hi Robert, thank you very much for the plots. Just a question. In your last plot, how did some genes in the "Moderated" t-test produce p-values < 1e-6 when the p-values of classical t-test were >= 1e4? To my eyes, the results of "moderated" t-test are actually more unstable for smaller p-value ...

written 14 months ago by
yjiangnan •

**10**0

votes

3

answers

2.2k

views

3

answers

... If the null hypothesis is true (e.g., compare two groups of samples from the same population), then the p-values of all genes surely would be uniformly distributed between 0 and 1. At the same time, given sufficient number of genes, the p-value of each gene approximates the proportion of genes more ...

written 14 months ago by
yjiangnan •

**10**0

votes

3

answers

2.2k

views

3

answers

... Sure, p-value is a random variable. But it has its own distribution and mean. When you have thousands of p-values of samples with the same mean, their mean would also converge to single value, which is what I referred to as a "true p-value". Because p-value has a uniform distribution between 0 and 1 ...

written 14 months ago by
yjiangnan •

**10**0

votes

3

answers

2.2k

views

3

answers

... Wikipedia definition: " p-value or probability value is the probability for a given statistical model that, when the null hypothesis is true, the statistical summary (such as the sample mean difference between two compared groups) would be the same as or more extreme than the actual observed results ...

written 14 months ago by
yjiangnan •

**10**0

votes

3

answers

2.2k

views

3

answers

... This hardly answers anything. Whatever methods you use, you should not go against the very definition of p-values. In the two experiments, after repeating millions of times, the true p-values are identical by definition. It is true that most of the time you could only estimate p-values because of th ...

written 14 months ago by
yjiangnan •

**10**#### Latest awards to yjiangnan

Popular Question
13 months ago,
created a question with more than 1,000 views.
For Are published RNA seq data analyses often wrong in calculating p-values and FDR?

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

Powered by Biostar
version 2.2.0

Traffic: 297 users visited in the last hour