help
help > RE: Please help: question about FDR correction
Jun 9, 2020 07:06 AM | Andrew Zalesky
RE: Please help: question about FDR correction
When using FDR, the number of permutations must be much much higher
than NBS.
For FDR, I suggest at least 50,000. 100,000 would better.
FDR results for 5000 permutations should NOT be used. They are NOT reliable.
Originally posted by Yongqin Xiong:
For FDR, I suggest at least 50,000. 100,000 would better.
FDR results for 5000 permutations should NOT be used. They are NOT reliable.
Originally posted by Yongqin Xiong:
Dear Andrew,
In the Manual NBS v1.2, it said when try using the FDR, it is important to increase the number of permutations, even 50000 permutations. Do I have to use 50000 permutations when using FDR? Why should I have to increase the number of permutations? If I get results when using 5000 or 1000 permutations, are these results reliable？ Can I report such results?
Thank you for helping this.
Yongqin
In the Manual NBS v1.2, it said when try using the FDR, it is important to increase the number of permutations, even 50000 permutations. Do I have to use 50000 permutations when using FDR? Why should I have to increase the number of permutations? If I get results when using 5000 or 1000 permutations, are these results reliable？ Can I report such results?
Thank you for helping this.
Yongqin
Threaded View
| Title | Author | Date |
|---|---|---|
| Yongqin Xiong | Jun 9, 2020 | |
| Yongqin Xiong | Jun 9, 2020 | |
| Andrew Zalesky | Jun 9, 2020 | |