Hi Reetu,
GARD is not designed to identify which sequences are recombinant. As we say in Multimedia File Viewing and Clickable Links are available for Registered Members only!! You need to
(page 8, this also applies to your large dataset question)
Quote:GARD is geared towards mapping the breakpoints and detecting segments of the alignment
which can be adequately described by a single tree topology; as we discuss in the next section, this
is necessary to allow more complex analyses to handle alignments with recombinant sequences.
Because GARD allows arbitrary tree changes across breakpoints, there are certain cases when it
will not perform well; for example, short alignments with many sequences. GARD requires about
approximately 4 times as many sites as sequences to run; otherwise the number of samples (sites)
is less than the number of model parameters (branch lengths and rates). Another case occurs when
only a few sequences in a large alignment have undergone recombination, in which instance the
cost of adding many new branch length parameters for one or more trees will likely outweigh the
likelihood improvement due to several local subtree rearrangements.
Generally, we recommend that you don't exclude recombinant sequences from downstream analyses,
but rather use those analyses which can handle recombinant sequences (see the link above for examples).
The fact that geneconv doesn't find any recombinant sequences isn't that unusual: it could reflect, for example,
ancient recombination events.
Sergei