Average Cline score differences for the benchmark tests of the DNA SMART dataset. Effect of ReformAlign to testsets of increasing APSI (A) and increasing number of sequences per alignment (B). The alignment pairs (from top to bottom, as appearing in the legends) correspond to [Ref. ClustalO – ClustalO], [Ref. ClustalW – ClustalW], [Ref. DialignTX – DialignTX], [Ref. GramAlign – GramAlign], [Ref. Kalign – Kalign], [Ref. Mafft (FFTnsi) – Mafft (FFTnsi)], [Ref. Mafft (Linsi) – Mafft (Linsi)], [Ref. Muscle – Muscle], [Ref. ProbConsRNA – ProbConsRNA] and [Ref. PicXAA – PicXAA]. For each value of the horizontal axis, the difference is calculated by subtracting the average Cline score of the reformed alignments to the respective score of the initial alignments that belong to each group. Statistically significant differences at the .05 significance level are denoted by an asterisk (*) at the top-left corner of the respective points.