Benchmarks against simulated unrelated protein and DNA sequences.

Large-scale random sequence tests indicate that for most alignment programs, aligned sequences are not necessarily homologous (table shows the fraction of random sequence aligned, calculated by taking a sum-of-pairs over pairwise alignments). Even when run in maximum-sensitivity mode (�Cmaxsn), FSA aligned only a small fraction of the random sequence. We generated 50 datasets, each with 10 random sequences, and ran all programs with default parameters. Protein sequences were 300 aa in length and DNA sequences were 1,000 nt in length. Results reported for ProbCons on DNA sequences were obtained with ProbConsRNA.