TRACS accurately classifies non-targeting controls and robustly classifies known essential and non-essential gene sets.

<p>(A) We evaluated the ability of MAGeCK to classify the 1,000 NTC sgRNAs in the GeCKO v2 pooled library as non-essential and compared it to TRACS as shown in <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0315923#pone.0315923.g002" target="_blank">Fig 2A</a>. The AUC for MAGeCK-RRA (51.3%) and MAGeCK-MLE (56.3%) were considerably lower than TRACS (98.5%). (B) We evaluated the ability of TRACS and MAGeCK to classify the previously described Hart <i>et al</i>. gene set of universally non-essential genes. TRACS (AUC: 93.5%) outperformed MAGeCK-RRA (AUC: 86.9%) and MAGeCK-MLE (AUC: 85.8%) suggesting it can reliably identify these non-essential genes. (C) We also evaluated the ability of TRACS and MAGeCK to classify a known set of universally essential genes. TRACS (AUC: 92.6%) consistently outperformed MAGeCK-RRA (AUC:78.1%) and MAGeCK-MLE (AUC: 85.9%) indicating it can robustly identify essential genes.</p> <p>(PDF)</p>