SNooPer: a machine learning-based method for somatic variant identification from low-pass next …
To achieve an accurate and unbiased estimate of the performance of SNooPer in predicting somatic variants, and to compare SNooPer to other routinely used somatic single nucleotide variant (SNV) callers including Varscan2 255 [27], JointSNVMix [29] and MuTect [31] (Additional file 2), we randomly excluded whole exome sequencing data from Dataset 1 before training and used it as test set (Fig. 2, Additional file 2). This test set is a particularly demanding dataset given its severely unbalanced class distribution, with approximately 1 true somatic variation per million false positives presenting at least one supporting read (TP/FP =9.3E-07).To accurately compare the performances of different algorithms, recall values were fixed for all callers and we estimated the precision (fraction of real calls) of each algorithm on the test dataset. Data were filtered on…
Link to Full Article: SNooPer: a machine learning-based method for somatic variant identification from low-pass next …
SNooPer: a machine learning-based method for somatic variant identification from low-pass next …
