Fig. 3
From: Detecting genomic deletions from high-throughput sequence data with unsupervised learning
Feature extractions from deletion candidates. Two deletion candidates are identified by discordant reads. “Deletion Candidate 2” is discarded after depth filter because its depth is larger than \(Depth_{avg}\). For “Deletion Candidate 1”, 5 ranges are identified by \(T_i\). \(L_i\) and \(D_i\) are the total length and the average depth of the range defined by \(T_i\) respectively. Each \(L_i\) is normalized by the length of “Deletion Candidate 1”, and the normalized results are recorded by \(LN_i\). Therefore, the internal structure of “Deletion Candidate 1” is presented by \(LN_{i}(i=0,1,2,3)\)