NMD prediction for stop_gained
EDIT: See VEP plugin NMD, which adds the term 'NMD_escaping_variant'.
Premature stop codons are more likely to be pathogenic if they cause nonsense-mediated decay (NMD). A common conception is that stop gain variants located <50 bp upstream of the last intron or in the last exon do not cause NMD (see Maquat 2004). However, using it the other way around should be done with care, as MacArthur 2012 states:
(...) most predicted NMD-triggering variants have no detectable effect on gene expression. \remark: 75 % (...) In addition, they demonstrate that the most widely used algorithm for NMD prediction (26) is an imperfect indicator of the effects of nonsense SNVs on RNA expression.
Therefore, NMD prediction should only be used as accessory information and not be implemented in set rules. One solution could be to display a warning "NMD not likely" for stop_gain variants (PVS1 candidates) located =<50 bp upstream of the last intron or in the last exon. Alternatively, stop_gain variants >50 bp upstream of the last intron could be marked with "NMD possible".
Original user request: LA-578