Table 2.

Suggested method (optimizes prediction accuracy and annotation error) vs. Read-length, where annotation accuracy = 100 - annotation error

Best Method (Prediction Accuracy/Annotation Accuracy)	100 bp	200 bp	300 bp	400 bp	500 bp	600 bp	700 bp
Single Method	MGA (57%/42%)	GM (77%/76%)	GM (86%/92%)	GM (87%/93%)	Orphelia (87%/95%)	Orphelia (89%/95%)	Orphelia (90%/94%)
Combined Method	Consensus (58%/47%)	Consensus (79%/84%)	Consensus (87%/96%)	Consensus (88%/96%)	GM&Orph (89%/96%)	GM&Orph (90%/98%)	GM&Orph (91%/97%)

The best method (for combined prediction accuracy and annotation accuracy, $\frac{predictio{n}_{-}accuracy+annotatio{n}_{-}accuracy}{2}$) vs. Read Length. The Method with Prediction Accuracy/Annotation Accuracy are in the parentheses. As shown, the Consensus gives the best overall performance for 100 bp-400 bp reads. For single methods, MGA has the best performance for 100 bp reads while GM is the best single-method for 200-400 bp length reads. For longer reads (500-700 bp), Orphelia and GM&Orphelia have the best performance.