Identification of CNV in the read-depth data. (A) CNV profile of a deletion locus in a read-depth data set. Short reads generated for a CEPH individual were mapped to the human reference genome and then counted in a 100-bp non-overlapping sliding window to produce the read-depth data. The counts are centered to their mean. The MCMC sampler was set up in a similar fashion as in the previous cases. The estimated means are plotted as a step function, and the 95% credible intervals as yellow boxes. The 2653-bp deletion, which is found by long sequence reads that encompass this deletion locus and thus split around it when aligned to the reference sequence, is taken as known and shown as the thin green line on a lower track. (B) The averaged standard deviation in the estimates of the start and the end positions (s and e) at different sequencing depths. The S.D. unit is the window size, which is 200 bp in this case.