Esp. with small assemblies of just a few reads, phrap has the nasty
habit of trimming off the first 30 bases or so off the ends if there
is only a single read's evidence at that position. Is there a way to
turn off trimming completely?
As our workaround, we actually duplicate the entries in the fasta file
while appending an "n" at the start and end of each copy ( or else
phrap will say it's a duplicate and toss it out ). This allieviates
the problem a lot, but not completely.