Fig. 5

Family-wise performance analysis on the Bralibase 2.1 pairwise benchmark set K2. (X-axis) The average difference between the Pankov and SPARSE for the sequence alignment quality metric SPS is computed per RNA family. (Y-axis) Similar to the X-axis but for the structure prediction metric MCC. Families with extreme differences are labeled. Pankov uses the superior estimation of the energy model while SPARSE applies PMcomp’s model (see Fig. 2). Pankov’s alignments gain in term of structure prediction for the majority of families