Next, we'll compute alignments using two structural alignment programs: LocARNA and r Coffee. LocARNA is an implementation of the Sankoff algorithm for simultaneous folding and alignment (i.e. it will generate both alignment and consensus structure. r Coffee uses an progressive alignment algorithm.
Download LocARNA from http://www.bioinf.uni-freiburg.de/Software/LocARNA/, extract and install it in your Progs folder and eventually add it to your path variable or copy it into the corresponding directory.
Both programs can read the fasta file four.seq.
$ mlocarna -p four.seq [...] K00283 GGGCCGGUAGCUCAUUUAGGCAGAGCGUCUGACUCUUAAUCAGACGGUCGCGUGUUCGAAUCGCGUCCGGCCCA K00349 GCCGAAAUAGCUCAGUUGGG-AGAGCGUUAGACUGAAGAUCUAAAGGUCCCCGGUUCAAUCCCGGGUUUCGGCA M10740 GCGGAUUUAGCUCAGUUGGG-AGAGCGCCAGACUGAAGAUUUGGAGGUCCUGUGUUCGAUCCACAGAAUUCGCA AF346993 CAGAGUGUAGCUUAAC---ACAAAGCACCCAACUUACACUUAGGAGAUU-UCAACUUAA-CUUGACCGCUCUGA alifold (((((((..((((.........)))).(((((.......))))).....(((((.......)))))))))))). \ (-52.38 = -21.42 + -30.96)
Use RNAalifold to predict structures for all your alignments (ClustalW, handcrafted, r Coffee, and LocARNA) and compare them. The handcrafted and LocARNA alignments should be essentially perfect, in the StrAl alignment the T-arm is slipped for AF346993.
Other interesting approaches to structural alignment include CMfinder, dynalign, and stemloc.