Gene Ranking of RNA-Seq Data via Discriminant Non-Negative Matrix Factorization (original) (raw)
Fig 3
AUC curve of differential expression analysis using ERCC and qPCR validated data.
(A) The cutoffs of absolute log fold change (LogFC) (0.58, 1 and 2) are used. DNMF works best in most cases. (B) At increasing LogFC, the performance of DNMF and gfold shows a slight advantage over other methods, while at lower logFC thresholds, DNMF generally outperformed other methods. AUC, area under the curve.