Skip to main content
Fig. 3 | BMC Medical Genomics

Fig. 3

From: Evaluating single-subject study methods for personal transcriptomic interpretations to advance precision medicine

Fig. 3

Exemplar accuracies of ss-DEGs methods validated using “All-against-One.” The selected seven ss-DEGs were evaluated against three rs-DEG-derived reference standards indicate a high-level of variability across ss-DEG methods and across reference standards, as well as a low to moderate-level of variability within ss-DEG methods and between biological replicates. The Precision-Recall and ROC curves across individual samples (Yeast) show that even in isogenic settings, a fair amount of biological variability exists. Furthermore, these single-subject studies provide a thorough comparison of each ss-DEG method’s performance and consistency in absence of replicates, allowing us to understand which tools have a greater potential for advancing precision medicine. For example, in the Yeast dataset, under > 40 biological replicates, the authors recommended DESeq and DESeq2. However, in absence of biological replicates, these techniques performed overly conservatively (unworkable recalls) and, on average, the worst

Back to article page