Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Differentiate DTU from DGE? #48

Closed
fruce-ki opened this issue Mar 1, 2018 · 0 comments
Closed

Differentiate DTU from DGE? #48

fruce-ki opened this issue Mar 1, 2018 · 0 comments
Assignees
Labels
enhancement Suggestions for better performance or better presentation.
Milestone

Comments

@fruce-ki
Copy link
Collaborator

fruce-ki commented Mar 1, 2018

Often, high-ranking DTU hits are just gene switch on/off events, where the very few transcripts seen in one of the two conditions can create an artificially massive change in isoform proportions.

Set better noise criteria, or find a way to flag and discount gene on/off events and marginal above-noise cases.
It is not RATs' intention to identify significant DGE in general, but when one of the conditions is very lowly expressed, it leads to results I do not find particularly convincing.

@fruce-ki fruce-ki added this to the 0.7.0 milestone Mar 1, 2018
@fruce-ki fruce-ki added the enhancement Suggestions for better performance or better presentation. label Mar 1, 2018
@fruce-ki fruce-ki self-assigned this Mar 1, 2018
@fruce-ki fruce-ki modified the milestones: 0.7.0, 0.6.3 Mar 8, 2018
fruce-ki added a commit that referenced this issue Mar 8, 2018
- 51: fixed the bug where accidentally the same isoform must exceed
noise in both conditions.
- 48: The noise threshold is applied to the total gene abundance as
well, preventing genes that are switched off in one condition from
creating wild proportions and exaggerated effect sizes.
@fruce-ki fruce-ki closed this as completed Mar 8, 2018
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement Suggestions for better performance or better presentation.
Projects
None yet
Development

No branches or pull requests

1 participant