On the probability that a novel variant is a disease-causing mutation

When a novel variant is found in a patient and not in a group of controls, it becomes a candidate for the disease-causing mutation in that patient. At present, no sampling theory exists for assessing the probability that the novel SNP might actually be a neutral variant. We have developed a population genetics-based method for calculating a P-value for a mutation-detection effort. Our method can be applied to a heterozygous patient, a homozygous patient, with or without inbreeding, or to a patient who is a compound heterozygote. Additionally, the method can be used to calculate the probability of finding a neutral variant at frequencies that differ between a group of patients and a group of controls, given some length of sequence examined. This method accounts for the multiple testing that is inherent in identification of variants through sequencing, to be used in subsequent case-control analyses. We show, for example, that for complete resequencing of 10 kb, the probability of finding a neutral variant in a patient and not in 50 controls is about 15%. Thus, discovery of a variant in a patient and not ill a group of controls is, on its own, very weak evidence of involvement with disease.