"alignment.hh" holds functions to:
- pre-compute the covariance of each two alignment columns (
covscore
)
- compute a representative character for an alignment column in terms of a "consensus" base, i.e. returns the most frequent base in this column. In case of co-optimals the alphabetically smaller base is reported. (
append_consensus
)
- compute a representative character for an alignment column in terms of a "most informative" base, i.e. bases in columns with frequency greater than the background
frequency are projected into IUPAC notation. Columns where gaps are
over-represented are in lower case. (
append_mis
)
- there is also the experimental filter function
maxcombsize
that should restrict the sum of both unpaired regions of an internal loop to a certain number of bases. This mimics size restrictions known from Vienna Package. Since we don't see a speed up, we stay with maximal 30 bases on each region individually.