 |
 |
Hi, The 0.5 means equal balance between false positives and false negatives. It doesn't mean 50% false positives or 50% false negatives. In fact, in an ideal dataset in which the separation between what is considered "signal" and what is considered "noise" is perfect, the 0.5 would lead to 0% false positives and 0% false negatives. Conversely, a poor separation between these would lead to a large number of both false positives and false negatives. If you really want 5% false positives, you need the mean and variance of the Gaussian that was fitted to the noise, and compute what is the z value for that distribution that yields p=0.05. I don't have any Melodic run right here with me to check the outputs but this information should be there among the outputs if the mixture modelling was applied. All the best, Anderson On 30 May 2018 at 21:48, huyang <[log in to unmask]> wrote: > Hi, Andreas > > Thank you for your help! That's indeed what I need. I read a paper ( > https://www.ncbi.nlm.nih.gov/pubmed/19357304) and found their threshold > is <0.05, but I don't know how to achieve it. I tested your solution and it > worked. But I just wonder whether there is any reference/documentation for > setting --mmthresh=0.05d? > > Best, > Yang Hu > > At 2018-05-31 01:40:47, "Andreas Bartsch" <[log in to unmask]> wrote: > > Hi, > > you can try --mmthresh=0.05d > for a local false discovery rate of 5% false positives but I am not sure > if this is want you want/need. > Cheers, > Andreas > > > Von: FSL - FMRIB's Software Library <[log in to unmask]> on behalf of > huyang <[log in to unmask]> > Antworten an: FSL - FMRIB's Software Library <[log in to unmask]> > Datum: Mittwoch, 30. Mai 2018 um 12:57 > An: <[log in to unmask]> > Betreff: [FSL] Mixture modelling threshold in melodic > > Dear FSL experts, > > After ICA decomposition, melodic would threshold the independent component > maps using mixture modelling at p = 0.5 by default, which would balance the > false positive and false negative. If I want to keep the false positive > rate to be lower than 0.05 (as we usually do using one-sample t-test), how > should I set the mixture modelling threshold? > > Best, > Yang Hu > > > > > ------------------------------ > > To unsubscribe from the FSL list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 > > > > > > ------------------------------ > > To unsubscribe from the FSL list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 > ######################################################################## To unsubscribe from the FSL list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1