We are trying to analyze some feature properties and sometimes there is a error in the processing of the FeatureFinderCentroided node. This error only happens with some samples which are not very different from samples in which the error does not occur. This is the message displayed:
, Abort reasons during feature construction:, - Could not extend seed: 44616, - Could not find good enough isotope pattern containing the seed: 11885, - Feature quality too low after fit: 5773, - Invalid feature after fit - too few traces or peaks left: 26943, - Invalid fit: Center outside of feature bounds: 1751, - Invalid fit: Fitted model is bigger than ‘max_rt_span’: 4617, - Invalid fit: Less than ‘min_rt_span’ left after fit: 12, FeatureFinderCentroided took 02:03 m (wall), 06:39 m (CPU), 31.27 s (system), 06:07 m (user).]
this indicates that maybe your settings are not robust enough. I think it is hard to “see” how similar two samples are for the underlying algorithm.
In general I would say that the output contains a bit too many features that are excluded so you should definitely have a look at the parameters.
They heavily depend on the type of instrument, the type of experiment and the type of molecules you are looking for (peptides? metabolites?)
If you are doing proteomics, you could also try the new FeatureFinderIdentification which focuses on features around MS2 Identifications.
If you are doing metabolomics, I strongly recommend FeatureFinderMetabo as it is the only one which is not focussed on peptides.
By the way, the error output seems to be truncated. Can you check that? In your output there is no Error but just warnings.
Hi,
The problem was solved either by using the FeatureFinderIdentification module or increasing the tolerance of the parameters of Feature FinderCentroid.
Thanks a lot,