Replies: 1 comment 1 reply
-
Hi @issararab, In mokapot we calculate the false discovery rate (FDR) for the PSMs accepted at a score threshold
Where The q-value is then a monotized version of these FDR estimates. Starting from the poorest score, we iterate backward through the list of scores. As we iterate, the q-value is the minimum FDR that has been observed. Below is an example to illustrate this. In the example, we consider higher scores as better:
When we perform a grouped confidence estimate, the q-values are merely calculated separately for each group denoted by the grouping variable. In practice, one must strike a balance when using this feature - the most benefit is gained when one chooses groups that should have different score distributions, but also not too small. If a group contains too few PSMs, then the Did that answer your question? |
Beta Was this translation helpful? Give feedback.
-
Hello,
I have a question about the group confidence score that mokapot computes. How does it compute the q-value?
I read through the paper that the training uses the whole set, but the grouped confidence score, which assigns the q-value, is calculated using the records within the group only.
isn't the q-value calculated as follows after sorting the new scores within the group?
q = (running decoys count) / (total running count)
Beta Was this translation helpful? Give feedback.
All reactions