v-annotate.pl: using -s with CMs built with --hand leads to alignment merging errors

[nawrockie]

For a CM built with --hand from an input alignment inalign, the resulting RF annotation when that CM is used to align sequences will be equal to the RF annotation in inalign, but v-annotate.pl -s expects it will be equal to the consensus sequence that is generated by the CM, and if it is not equal alignment merging will usually fail and the script will exit.

So, currently, if you want to use such a CM, you need to either manipulate the RF annotation within the CM file to match the CM consensus sequence (output with cmemit -c) or rebuild the CM using a modified inalign file that has the RF annotation redefined as equal to the consensus sequence.

[nawrockie]

How to check if a CM being used with v-annotate.pl has this problem:

> cmemit -c my.cm | grep -v ^\> | sed 's/\(.\)/\1\n/g' | grep . > consensus.txt
> cmemit -c my.cm | cmalign --outformat pfam my.cm - | grep RF | awk '{ print $3 }' | sed 's/\(.\)/\1\n/g' | grep . > rf.txt 
> wc -l consensus.txt
7240
> wc -l rf.txt
7240
> diff consensus.txt rf.txt

Any RF characters in rf.txt that don't already match should be changed to match (including lower/uppercase) the corresponding position in the consensus in consensus.txt