Who can explain extrinsic.M.RM.E.W.cfg column 2-4 in [GENERAL] section

[xiekunwhy]

Hi,

Why extrinsic.M.RM.E.W.cfg has one more column before extrinsic columns, and what column 2-4 means?
Why other extrinsic.cfg files have one column less than extrinsic.M.RM.E.W.cfg before extrinsic columns?

Also appear here (https://vcru.wisc.edu/simonlab/bioinformatics/programs/augustus/docs/tutorial2015/prediction.html#extr)

Best,
Kun

[MarioStanke]

These numbers are maluses (pentalties) for predicting gene features that are not supported by hints.
The exonpart malus of .997 is multiplied for each candidate exon base that is not covered by any exonpart hint.

In the picture there is a 'local' CDS part malus of 0.985. This applies only to bases of a candidate CDS that otherwise have evidence, i.e. the code checks that there are hints for parts of the CDS and no hints for other parts.
It is intended to penalize extra/unsupported parts of an exon that is supported by evidence elsewhere. A regular malus like the exonpart malus of .997 here applies to any base without hints and does.