Multiple alignment program for amino acid or nucleotide sequences

Supported in version 7.034 and higher (2013/Apr/24).
Updated in version 7.043 (2013/May/26).
Updated in version 7.236 (2015/May/31).
This feature was unstable due to a bug in versions 7.307 (2017/Jan) – 7.396.  Fixed in 7.402 (2018/May/23).

Merge multiple sub-MSAs into a single MSA

Online version In alpha testing (2015/Jun)

Command-line version

Two or more sub-MSAs (and unaligned sequences) can be merged into a single MSA by the --merge option.  Each sub-MSA is preserved.

1. Concatenate the multiple sub-MSAs (in multi-fasta format) into a single input file.

   % cat subMSA1 subMSA2 subMSA3 otherSequences > input
   

   Updated (2020/May/14) Note that each file must have "end of line" at the end of file.  If not sure, type:

   % echo >> subMSA1
   % echo >> subMSA2
   % echo >> subMSA3
   % echo >> otherSequences

   before running cat. 
2. Create a file (say subMSAtable) to specify which sequences correspond to sub-MSAs, using the makemergetable.rb script.

   % ruby makemergetable.rb subMSA1 subMSA2 subMSA3 > subMSAtable
   

   The subMSAtable file is:

   1 2 3 4 5  # this is comment. subMSA1
   6 7 8      # you can write anything after #
   9 10       # subMSA3
   

   This means:
   • Sequences 1-5 are already aligned.
   • Sequences 6-8 are also already aligned.
   • Sequences 9 and 10 are also already aligned.
   • Sequences that are not listed here (if any) will be ungapped and then aligned.
   Sequences are represented as numbers (1, 2, 3 ..), according to the order of occurrence in the input file.

   The subMSAtable file can be manually edited as necessary.

3. Run

   % mafft --merge subMSAtable input > output
   

   A more rigorous distance measure can be used for small data.

   % mafft --localpair --merge subMSAtable input > output
   

   Iterative refinement is supported in version 7.040 and higher.

   % mafft --localpair --maxiterate 100 --merge subMSAtable input > output
   

Each sub-MSA is forecd to form a monophyletic cluster in version 7.043 and higher (2013/May/26).

If phylogenetically unnatural grouping (eg, "1 5" in the above case) is given, the following warning appears and the calculation becomes slow.

# Sequences 1 and 2 seem to be closely related, but are not in the same sub MSA (1) in your setting.

This warning has been changed in v7.236 (2015/Jun).

If inconsistent groupings (eg, "1 2" and "1 3") are given, the program stops.

By the --treein option, a user-defined guide tree can be given.  In this case, each sub-MSA must form a monophyletic cluster in the given guide tree.  Otherwise, the program stops.

% mafft --treein tree --merge subMSAtable input > output

The --merge option completely covers the situations where the mafft-profile program was used.  So, mafft-profile will be deleted in the future.

% mafft-profile subMSA1 subMSA2 > output

   ↓↓↓

% cat subMSA1 subMSA2 > input
% ruby makemergetable.rb subMSA1 subMSA2 > subMSAtable
% mafft --merge subMSAtable input > output

Bug information (2013/May/26):
In the multithread mode (--thread #), the --merge feature did not work in version 7.040.  This bug has been fixed in version 7.043.

Combination of --merge and --seed

The --merge option can be combined with the --seed option.

% mafft --localpair --seed seed --merge subMSAtable input > output

The sequences in the seed file are first numbered (1, 2, 3, ..), and the sequences in the input file are numbered (4, 5, 6, 7 ...).  This numbering must be used in the subMSAtable and tree files.

To generate subMSAtable in this case, use the -s option to specify the number of sequences in the seed, eg, the seed file has three sequences,

% ruby makemergetable.rb -s 3 subMSA1 subMSA2 > subMSAtable

The resulting subMSAtable is:

4 5 6 7 8  # subMSA1
9 10 11    # subMSA2

This can be used for the above command.