Figure 5.

Accuracy vs. Fragment Types for Each Method. The accuracy for different fragment types vs. several Boolean logical combinations of the algorithms. All reads lengths for each type were averaged, resulting in an average of 7000 simulated reads for each fragment type. We can see that GM&Orph significantly improves accuracy of Type D fragments, GM|MGA|Orph improves the gene edges, while the consensus Boolean combination marginally improves all fragment types.