In QPX; each feature is "disconnected" from the PSMs; when you are at the feature level, you can't directly to the PSM table and check all the PSMs for a given feature; we can infer that by knowing the retention time of the psm + the peptidoform and look if they are in the range of start and end of RT of the feature.
We are tracing now only the best PSM that was used to identify the feature, even with MBR, but it would be nice, at least for non-MBR, to trace and hardcode for every feature all the PSMs that belong to it. It could be like a list of scans ids which rigth now is also a list.
The MBR features is more complex because the ids could come from different places; then we could if we want to generalize more doing like:
id_scans = struct{
run_file_name = ""
scans = [
[]
]
}
This could give better provenance. What do you think @jpfeuffer @timosachsenberg
In QPX; each feature is "disconnected" from the PSMs; when you are at the feature level, you can't directly to the PSM table and check all the PSMs for a given feature; we can infer that by knowing the
retention time of the psm+the peptidoformand look if they are in the range of start and end of RT of the feature.We are tracing now only the best PSM that was used to identify the feature, even with MBR, but it would be nice, at least for non-MBR, to trace and hardcode for every feature all the PSMs that belong to it. It could be like a list of scans ids which rigth now is also a list.
The MBR features is more complex because the ids could come from different places; then we could if we want to generalize more doing like:
This could give better provenance. What do you think @jpfeuffer @timosachsenberg