Acceptable protein group probability and FDR settings

[Vidya Venkatraman]

Hi,

I have a question about acceptable protein group probability and FDR settings.

Is it acceptable to use protein hits with lower protein group probability (eg 50%) if it gives an estimated FDR (as calculated by looking at the decoys) of ~1%? If it is acceptable, why do all the papers simply report protein group probability >= 90%?

Is it not acceptable to report things from an FDR perspective and then state that the score cutoff (ie the probabilty cutoff) to achieve this was 50%? Or does "50%" mean that half of the protein are probably not true (which does not seem accurate based on the decoy FDR estimations....)? 

Can we rely on such proteins (with 1%FDR cutoff but low protein group/low peptide probability (eg. 50%)) to be a true hit from the spectrum standpoint?

Any input on acceptability/reliability of low probability hits (with 1% FDR cutoff) are welcome.

Thanks

Vidya