An amino acid data set of 14,382 positions was prepared as follows: the deduced amino acid sequences from the 71 individual genes were aligned using MUSCLE 3.7 [ 72], the ambiguously aligned regions in each alignment were removed using TRIMAL 1.3 [ 73] with the options block = 7, gt = 0.7, st = 0.001 and sw = 3, and the protein alignments were concatenated.
This data set was prepared as follows: the deduced amino acid sequences from the 69 individual genes were aligned using MUSCLE 3.7 [ 57], the ambiguously aligned regions in each alignment were removed using TRIMAL 1.3[5858] with the options block = 6, gt = 0.7, st = 0.005 and sw = 3, and the protein alignments were concatenated using Phyutility 2.2.6 [ 59].
The ambiguously aligned regions in each alignment were removed and optimized using GBLOCKS 0.91b [ 89] with the -b2 option (minimal number of sequences for a flank position) set to 13.
The poorly aligned positions and divergent regions in each alignment were eliminated with GBlocks program [ 42] and the resulting blocks were concatenated in a single alignment of 15195 positions.
The AVANA tool was used to select nonamers with conservation of ≥80% in each alignment.
Thus, the number of sequences in each alignment ranges from 4 to 807.
Sequences in each alignment were evaluated for overlap and sequence identity.
We removed all exonic sequences from the alignments and then randomized non-gapped columns in each alignment.
A maximum of two mismatches were allowed for first 32 bases in each alignment, and reads that had more than two mismatches or were mapped to multiple genomic locations (alignment score less than 4) were discarded.
The method exploits information from the reference alignments to classify the sequences in each alignment into a number of subfamilies and to construct a representative model for each protein subfamily, including characteristic conserved blocks and typical start/stop sites.
For all analyses, we used the Bowtie [7] program and the same parameters to align these reads (see Materials and Methods for details), in which we allowed up to three mismatches in each alignment.
