This is the pfold RNA fold server. It takes an alignment of
RNA sequences as input and predicts a common structure for all
sequences. The algorithm is described in the references below.
A new version of the algorithm has been created, called PPfold.
PPfold is a parallelized version of pfold, and can predict the
structure of much longer alignments without underflow errors. You
can find out more about PPfold here. This webservice still uses the old version (pfold). We
are working on creating a webservice for PPfold, which will
eventually replace pfold.
If you experience any issues, or have comments, email:
Zsuzsanna Sukosd (email@example.com)
Insert your sequences in fasta format in
the box below. An example of the fasta format is shown in the box. At
the moment, the server has a maximum of 40 sequences, and an alignment
length of 500.
The accepted nucleotide symbols are: 'A', 'B', 'C', 'D', 'G', 'K',
'M', 'N', 'R', 'S', 'T', 'U', 'V', 'W', 'X' and 'Y', as used in the
extended nucleotide alphabet. Gap signs, '-', are also accepted along
with the above letters in lower case.
When you press `Fold RNA' your data will be processed.
Pfold was originally written by Bjarne Knudsen.
Known structure can be incorporated by including a sequence called 'structure'. The structure is indicated in three ways:
- By parentheses. Columns with matching parentheses are forced to pair with each other
- Using 's'. Columns with an 's' will be forced to be single stranded.
- Using 'd'. Columns with a 'd' will be forced to pair with another column.
All symbols different from the above means that nothing is assumed
about the structure of that position. An
example. Notice that some structures are impossible and may yield
strange results (e.g. all positions paired and loops of size less than
The rate matrices and grammar of the 1999 paper are given here:
Unpaired rate matrix
Paired rate matrix