We set a strict threshold for the Mendel violation p value of 10−9 such that in 500 trios, we expected less than one false positive. As previously indicated, we also set a strict threshold for the population filter of no more than five parents showing a lesion involving a given probe. This method identified 70 de novo copy-number events in 67 trios. We performed manual curation, in which

we relaxed the p value threshold to 10−7 and the population threshold to 20. This yielded 241 de novo candidate (DNC) events in 216 children. For each DNC, we assessed a variety of selleck products information such as family ratio data, modeled state means, population polymorphism, quantile quality scores, and systematic noise. A total of 91 events passed curation, including all 70 stringent events. A full list of de novo events and their method of discovery can be found in Table S1. Given the limited size of the X and Y chromosomes, we chose not to automate de novo discovery over these chromosomes. We altered the five-state model to use a reference Docetaxel nmr copy-number state of 1 and modified the Mendel violation rules for a probe to reflect the gender of the child and the parents. We then manually inspected all segments with greater than 70% of the probes reporting

as Mendel violators. Using this method, we identified three X chromosome de novo events (Table S1). To identify transmitted copy-number events, we developed a 125-state HMM that operates simultaneously on the normalized ratio data of the child, father, and mother. To determine emission probabilities, we used the product of the

five-state model for each member of the trio. We limited the effect of isolated failed probes by setting a minimum emission probability calibrated to the rate of single probe outliers. Transition probabilities were computed from the average CNV frequency based on KS segmentation. An additional penalty was applied for entering a “Mendel-violating” state. We then employed the Viterbi algorithm to find the most likely path through the state space. Restricting to events in which the child showed deletions or duplications, we then determined whether any Isotretinoin parent shared the event. For each of the eight possibilities (del/dup; from mother/father/both/neither), we constructed a measure of support similar to that of Mendel violators. Worst-case false-positive rates were determined and p values assigned to each transmitted (and de novo) event using a binomial distribution. To determine the statistical significance of asymmetries, we performed random permutations of the data. Typically, we used 10,000 permutations for each test. See Supplemental Experimental Procedures for more details. This work was supported by a grant from the Simons Foundation (SFARI award number SF51 to MW).