Hi Lars  and Andreas,

In large protein groups this is expected behavior since the peptides are
highly degenerate and their weights drop below the 0.5 cutoff required to
compute a probability for any one given protein.  There is an option for
ProteinProphet called GROUPWTS that will use the sum of the weights among
all proteins within the group for a given peptide to determine if the
peptide will be used to compute a probability.  With this option your
grouped proteins should now have a probability to allow a secondary ranking
among proteins within a group.

-David

On Thu, Oct 25, 2012 at 7:30 AM, Andreas Quandt <[email protected]>wrote:

> Dear TPP developers,
>
> I might have uncovered a bug in ProteinProphet, or at least stumbled
> across a result which does not make sense to me. We are using all annotated
> isoforms from Swissprot when we search data from human. This roughly
> doubles the number of human entries from 20k to 40k many which as identical
> over long stretches. It seems I end up with protein groups of high
> probability but all members have a probability of 0. Is this expected:
>
> <protein_group group_number="311" pseudo_name="50" probability="1.0000">
> <protein protein_name="P05164" n_indistinguishable_proteins="**5"
> probability="0.0000" ...
> ...
> <protein protein_name="P05164_CHAIN_2" n_indistinguishable_proteins="**1"
> probability="0.0000" ...
> ...
> <protein protein_name="P05164_CHAIN_3" n_indistinguishable_proteins="**1"
> probability="0.0000" ...
> ...
> <protein protein_name="P05164_CHAIN_4" n_indistinguishable_proteins="**1"
> probability="0.0000" ...
> ...
> </protein_group>
>
> The result for the same data, same search parameters but with a database
> not containing isoforms I get:
>
> <protein_group group_number="26" probability="1.0000"> <protein
> protein_name="P05164" n_indistinguishable_proteins="**1"
> probability="1.0000" percent_coverage="27.7"
> ...
> </protein_group>
>
> I would greatly appreciate if could tell me if this behavior is expected.
>
> Thanks!
>
> Lars Malmstroem
> ETH Zurich
> Switzerland
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"spctools-discuss" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/spctools-discuss?hl=en.

Reply via email to