[MUMmer-help] nucmer misses alignments with less input sequences

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

Hi,

I am using nucmer to align DNA sequences and during some playing around 
with my input data, I noticed the following bug or strange behavior:

I aligned two small files (input.fa with sequence "a" and "b" against 
input2.fa with sequence "c") and found a hit between "a" and "c"

cat input.fa
 >a
GTGTCTTTGGTGCACTGGATGTAGATGGATTCTTGCGCCCATTCACGATTGTGGTAGAAGGGCTAGTAAAGGTGGAGTGGATAATCCAACCCGCAACTCCGATATGAATGGCCGTCTTTTGATCAGCCGCATATTAGACTTAACGTGCTCAAAAACACGGTTCATAATCCTTGATTATGCTGTTGACCAGCTTGCTGGCCAGTGTAGGATCAATTCGTTTATTATATAAGTCCTGCAGTCAAAGATCAGTAAGAAATTATTGATTGCTTTCTAACGTTTTTTCCCTTTCATGAAGTGCTTGGGGTTTTGCCAAAGAAATCCTTTAGCCTTTAGTAGTAAATAAGCAGACTAGAGTCCACATACATCGACAATATATCAGCTTCTGCTGGCGGCGAATCAGAGGGGCCGCTGTCGCCTCCACCTATGAGGCCCCTCAGGGCGAGCTGGCTCCATTGGGAGTAGATGGAACACTGGCGGCTTCAGTGGGTTAGACCTGTGTAATAGAGACACGCCAGATTTGCACCCTGCCCCTGGTGGTGCACTGTCGCCAATCGCCCAAGCGTGGCGACACCTTGATCCTGGCCTGTTCCCGATTGGGCAATCGGGCCTGGGGGGCAGTCGTTCCAGGCCCAAATTAGCTGCATTGCATCGTGTGGCATCACTGGCGCACCTGCAGTGGCGCCTCCGGCCCTCGCCTTCCTGCTCCTACTGTGTCAGCCAGTGCATTCACGCATCTATCCATAGAGATACATTTGGGTGTGCCTGGGCTTTCGCACCAGCGATTAACTTTTTGCTGCCAGTTTCCTCCTTGGCTTACTGCAGGCGGCAAGATGGAACAGCAGAATGTCAGCTGACATTCGTTCACAAAAGATCATGAAACTAACACCAGGGCCGGAATTGTAGAGATGTATCCACTTCCTTGTGTGTTTAGGAACTAACACAAGTTGGTGGATGGGTGGTAGTAACAATGACAGAAGAATTGCCACCCGG
 >b
AAAAGAGAGCATTTATGCTTTTGAATTTAACTTAAGTTTGATGCACTAAAATAATAAAGGTATTCGATGTACTGTTCTATTTAAGTGCATAAAACTATTTTTTGTAAGGGTGCTATATCCTTTTTTCAGATAGATTCTTTGTTGATTCTCTTCCTTTTGATAAGAAGGAGTAGTTCCTGGTCAAGAGGCTAAAGTGTGGAAATGACCACATTTTCCACACTTTCTGCCAACCAGTGACAACATATCATAGGTGGCTAGCAGTGCAGAAAGAAGTCAAAAAGGTGTCCTGGCTTCAGGTTAGCCTCCTCCAGCCTAGCCAACAAAATGTGACTGCTTTGAATTCTCTGTAAACAAGAGAATTCATAAAAGCCATAAAACCAAAAAAAAAAACTCTGCACACTTCCAAGCAGGAGCTGTAGCTGGCAGATGTAGATCTAAGTTGGGTTTTATCCAGCTAGGGCCAATTATCCTCTCCTGTTCTGTTAGCCAAATTGTGGGATTCTCTAAAAGCCCAGAAATTACTCGCTGATAGAGTTGGAAACACCTAGACAAAGGGCTAAAAAGTTCAGGAGCATTAAAAAATGGTATTCCTGCATTGAACAGTACTCGCCCTCCAAAGAATTCATCTATTTGAAGGTTAAAAATGTGGGATCAAATATAAAAGTAGATCCTTAACCCAATACCAATGTGTGTTTACGCAACATTAAGGCTTTTATTTACAAGATTTGGGAGAAGAGAAGATGGATGAAGGCTGTTGAGTAAAATAATGAAACAGATGAATTTTTCCCCTTGAAAACAGAGACACAGGTTAAACCAGGAGGGGTTGGAGAATAGCCAATGGTTAACATGTATGCCCAACTGATGAGATGTTTTTAAAGATGATAAAGAGAAGGAGGGTTCCTTCCTGGTTGAGTAGAAACGGAGGGTGGCTTTTTTGAAGAGGAGCGAAA
cat input2.fa
 >c
TATAAATAGCTAAATACATAACTAATGTTATGAATTCTTCTAGAAAATGTCAAGAAGTCATTGCAATTTATTTCTCGCTTGCAACTTTGAAACAATATTGTATGCTTTAGGTTCCACTTTCCACTTTTTTCTTGCCTCTGTCTGGTTGATTATTTTTGTGTCAGTGGTGCCACCTCGTGGAGTTTAAGTTTTACTGGCAAACTCTCCTTTTAAAATAAACACAGTGTAGTATTATTATATGGGAAAGAGAACAGGATTCTTGTGCATTCAGTAGTTATACAGAACACCTGCTTAAATTTTGTCTTTGGTACAACTATTAGAAGCACAGAAGTCCTTCTAACTGTACTAAAGTGGATCATTGCAAAGTCATTATGAAAAGAATTCTAACAAAACGCATACATTGCAACAGAATGAAAACATGGCAATTCTTGTATTTTATAACCCATCCAAAACTTGTGTTAGTTCCTAAACACAAGGAATGGAGTACATCTACAGTTCTGTTAGTTCATGATTTTTGTGAACATGTCAGTGACATTCTGCTGTTCCATCTTGCCGCCTGCAGTAGCCAAGAGAAGTGGCAGCAAAAAGTTAATCACTGGTGCAAACCCAGCAACCAAATTATCTCTATGGATAATGGTGAAGGCACTGGCTGACACAGTAGGAGATGAAGGCGAGGCGAGGCGAGGCCATGCAGTGCGCAGTGATGCACAGATGCAATCAGCTAATTGGGCCTGAACGATGCCCCAGGCCCGATTGCCCAATCGAACAGGCAGGATCAATGTGTCGCAGTGGCCGGGACA
nucmer --maxmatch -l 11 -b 1000 -g 1000 input.fa input2.fa ; show-coords 
out.delta

gives:
NUCMER

     [S1]     [E1]  |     [S2]     [E2]  |  [LEN 1]  [LEN 2]  |  [% 
IDY]  | [TAGS]
=====================================================================================
      565      985  |      787      420  |      421 368  |    83.89  | 
a    c

Now, if I remove sequence "b" from input.fa that shows no alignment, the 
alignment between "a" and "c" is not found anymore

cat input.fa
 >a
GTGTCTTTGGTGCACTGGATGTAGATGGATTCTTGCGCCCATTCACGATTGTGGTAGAAGGGCTAGTAAAGGTGGAGTGGATAATCCAACCCGCAACTCCGATATGAATGGCCGTCTTTTGATCAGCCGCATATTAGACTTAACGTGCTCAAAAACACGGTTCATAATCCTTGATTATGCTGTTGACCAGCTTGCTGGCCAGTGTAGGATCAATTCGTTTATTATATAAGTCCTGCAGTCAAAGATCAGTAAGAAATTATTGATTGCTTTCTAACGTTTTTTCCCTTTCATGAAGTGCTTGGGGTTTTGCCAAAGAAATCCTTTAGCCTTTAGTAGTAAATAAGCAGACTAGAGTCCACATACATCGACAATATATCAGCTTCTGCTGGCGGCGAATCAGAGGGGCCGCTGTCGCCTCCACCTATGAGGCCCCTCAGGGCGAGCTGGCTCCATTGGGAGTAGATGGAACACTGGCGGCTTCAGTGGGTTAGACCTGTGTAATAGAGACACGCCAGATTTGCACCCTGCCCCTGGTGGTGCACTGTCGCCAATCGCCCAAGCGTGGCGACACCTTGATCCTGGCCTGTTCCCGATTGGGCAATCGGGCCTGGGGGGCAGTCGTTCCAGGCCCAAATTAGCTGCATTGCATCGTGTGGCATCACTGGCGCACCTGCAGTGGCGCCTCCGGCCCTCGCCTTCCTGCTCCTACTGTGTCAGCCAGTGCATTCACGCATCTATCCATAGAGATACATTTGGGTGTGCCTGGGCTTTCGCACCAGCGATTAACTTTTTGCTGCCAGTTTCCTCCTTGGCTTACTGCAGGCGGCAAGATGGAACAGCAGAATGTCAGCTGACATTCGTTCACAAAAGATCATGAAACTAACACCAGGGCCGGAATTGTAGAGATGTATCCACTTCCTTGTGTGTTTAGGAACTAACACAAGTTGGTGGATGGGTGGTAGTAACAATGACAGAAGAATTGCCACCCGG
cat input2.fa
 >c
TATAAATAGCTAAATACATAACTAATGTTATGAATTCTTCTAGAAAATGTCAAGAAGTCATTGCAATTTATTTCTCGCTTGCAACTTTGAAACAATATTGTATGCTTTAGGTTCCACTTTCCACTTTTTTCTTGCCTCTGTCTGGTTGATTATTTTTGTGTCAGTGGTGCCACCTCGTGGAGTTTAAGTTTTACTGGCAAACTCTCCTTTTAAAATAAACACAGTGTAGTATTATTATATGGGAAAGAGAACAGGATTCTTGTGCATTCAGTAGTTATACAGAACACCTGCTTAAATTTTGTCTTTGGTACAACTATTAGAAGCACAGAAGTCCTTCTAACTGTACTAAAGTGGATCATTGCAAAGTCATTATGAAAAGAATTCTAACAAAACGCATACATTGCAACAGAATGAAAACATGGCAATTCTTGTATTTTATAACCCATCCAAAACTTGTGTTAGTTCCTAAACACAAGGAATGGAGTACATCTACAGTTCTGTTAGTTCATGATTTTTGTGAACATGTCAGTGACATTCTGCTGTTCCATCTTGCCGCCTGCAGTAGCCAAGAGAAGTGGCAGCAAAAAGTTAATCACTGGTGCAAACCCAGCAACCAAATTATCTCTATGGATAATGGTGAAGGCACTGGCTGACACAGTAGGAGATGAAGGCGAGGCGAGGCGAGGCCATGCAGTGCGCAGTGATGCACAGATGCAATCAGCTAATTGGGCCTGAACGATGCCCCAGGCCCGATTGCCCAATCGAACAGGCAGGATCAATGTGTCGCAGTGGCCGGGACA
nucmer --maxmatch -l 11 -b 1000 -g 1000 input.fa input2.fa ; show-coords 
out.delta

gives:
NUCMER

     [S1]     [E1]  |     [S2]     [E2]  |  [LEN 1]  [LEN 2]  |  [% 
IDY]  | [TAGS]
=====================================================================================

I called mummer and mgaps manually on the .ntref file and it seems that 
while mummer outputs something, mgaps removes the alignment.

Could you please have a look if this is a bug, and if not tell me which 
mgaps parameters would keep that alignment?

Thanks a lot
- Michael