Table 1. Accuracies for Calling C5-Cytosine Variants Using a Random Forest Classifier^a.

context	count	accuracy
AnCGA	250	96.6 (±0.54)
AnCGC	250	95.9 (±0.72)
AnCGG	251	91.9 (±0.57)
AnCGT	250	92.7 (±0.93)
CnCGA	287	94.0 (±0.55)
CnCGC	251	97.0 (±0.36)
CnCGG	330	93.3 (±0.61)
CnCGT	250	93.4 (±0.94)
GnCGA	250	93.6 (±0.75)
GnCGC	513	98.3 (±0.22)
GnCGG	250	94.2 (±0.69)
GnCGT	250	96.5 (±0.65)
TnCGA	250	95.6 (±0.81)
TnCGC	250	95.0 (±0.66)
TnCGG	250	95.6 (±0.61)
TnCGT	259	91.6 (±1.19)

^aTwenty iterations of 5-fold cross-validation were performed for each XnCGY context. For each iteration, an accuracy measurement was made (% correct across entire data set of C, mC, hmC, fC, and caC for that context). Column 3 is the mean and standard deviation of those 20 measurements. Column 2 is the total number of events quantified for each XnCGY context.