When to use the activating vs deactivating predictor?

This predictor represents a typical situation when one has what is believed to be a functional variant (e.g. observed many times in a cohort or dataset) and wishes to distinguish these two possibilities. For example:

The BRAF p.Val600Glu variant
  • Val600 in BRAF lies within the activation-loop
  • Val600 is the position within COSMIC with the most missense variants (53122).
  • BRAF p.Val600Glu represents a variant you'd expect to have a functional consequence on kinase activity of BRAF.
  • Val600Glu is predicted activating (see right).
    • workflow

    The STK11 p.Ser216Phe variant
  • Ser216 lies within the activation-loop of STK11's kinase domain
  • Ser216 is the position within COSMIC with the most missense variants (21) in the kinase domain.
  • STK11 p.Ser216Phe represents a variant you'd expect to have a functional consequence on kinase activity of STK11.
  • Ser216Phe is predicted deactivating (see right).
    • workflow

    When to use the activating vs deactivating vs neutral predictor?

    This predictor represents is more reflective of a situation where one does not know if a variant is functional at all and thus one needs to predict neutrals. For example:

    The MELK p.Lys219Arg variant
  • Lys219 lies in the αG helix of MELK
  • Lys219 has a minor allele frequency of 2%.
  • MELK p.Lys219Arg represents a variant you wouldn't know if it has a functional consequence on kinase activity of MELK.
  • Lys219Arg is predicted neutral (see right).
    • workflow

    When to use the resistant or not predictor?

    This predictor predicts if the given variant is resistant or not. However, it does not take into account if the variant is activating or deactivating. Hence, this predictor is useful when one knows that a variant DOES NOT lead to loss of kinase activity but one does not know if the variant is resistant or not (see example below). For example:

    The EGFR p.Thr790Met variant
  • Thr790 lies in the β5 sheet of EGFR
  • Thr790Met is predicted to be drug-resistant (see right).
  • Note that this predictor predcits the likelihood of a given variant to be resistant to a drug, not the likelihood of a patient to respond to a drug.
    • workflow

    The RAF1 p.Asp486Asn variant (CAUTION)
  • Asp486 lies in the activation-loop and at the DFG-motif in RAF1
  • Asp486 is predicted to be drug-resistant (see right-most figure).
  • Asp486 is also predicted to be strongly deactivating (see left-most and middle figures)
  • Since the resistance or not predictor lacks the information about activating or deactivating, it is not able to take this into account and hence the prediction of drug-resistance should be ignored.
    • workflow
    workflow
    workflow