77
5. Select “BIONJ tree” as the base tree for likelihood calculations (see Note 9).
6. After the computation is complete, select “Results” under the
Selection menu. Identify the model with the lowest AIC,
which will be used for the subsequent analysis.
Given an alignment of protein sequences and a probabilistic model
of protein evolution, a phylogenetic tree can be inferred using the
maximum-likelihood method. The goal of phylogenetic inference
using ML is to identify the tree topology (branching pattern) and
parameters (for example, branch lengths) that optimize the likelihood function. Likelihood is defined in this context as the probability of observing the data—that is, the probability that the
present-day sequences would have evolved—given the tree topology, tree parameters, and an evolutionary model. In practice, the
maximum-likelihood tree is inferred by first estimating the correct
tree using a computationally simple method such as the BIONJ
method; heuristic tree-searching algorithms are then applied to
modify the topology of the tree and increase its likelihood until
convergence on the maximum-likelihood tree is achieved.
Inference of a phylogenetic tree for the purpose of APR should
be considered an iterative process; problematic sequences should
be identified and removed, and alternative sequences should be
added before repeating the phylogenetic analysis. The final tree
should be biologically plausible, supported by high bootstrap values, and robust to variations in the evolutionary model used.
1. Open the PhyML [25] web interface and upload the sequence
alignment.
2. Select “Amino-Acids” for data type.
3. Under “Substitution Model,” specify the substitution model
chosen in Subheading 3.3. +I denotes an estimated proportion of invariant sites; +G denotes an estimated gamma-shape
parameter, and +F denotes empirical amino acid frequencies.
4. Under “Tree Searching,” select “BIONJ” for the starting tree and
“NNI + SPR” for the type of tree improvement (see Note 10).
5. Under “Branch Support,” select the “aLRT SH-like” branch
support test. Alternatively, for the final tree, select a bootstrap
calculation with 100 replicates.
6. Once the analysis has completed, open the maximum-likelihood tree in FigTree. The tree can be rooted by selecting the
branch connecting the ingroup and outgroup sequences and
clicking the “Reroot” button.
7. Inspect the tree to identify problems such as long branches
and low branch support values, and rectify these problems by
repeating the multiple sequence alignment and phylogenetic
analysis using a modified sequence dataset if necessary
(see Notes 11 and 12).
3.4 Phylogenetic
Tree Inference
Improving FRET Sensors by Ancestral Gene Resurrection
5. Select “BIONJ tree” as the base tree for likelihood calculations (see Note 9).
6. After the computation is complete, select “Results” under the
Selection menu. Identify the model with the lowest AIC,
which will be used for the subsequent analysis.
Given an alignment of protein sequences and a probabilistic model
of protein evolution, a phylogenetic tree can be inferred using the
maximum-likelihood method. The goal of phylogenetic inference
using ML is to identify the tree topology (branching pattern) and
parameters (for example, branch lengths) that optimize the likelihood function. Likelihood is defined in this context as the probability of observing the data—that is, the probability that the
present-day sequences would have evolved—given the tree topology, tree parameters, and an evolutionary model. In practice, the
maximum-likelihood tree is inferred by first estimating the correct
tree using a computationally simple method such as the BIONJ
method; heuristic tree-searching algorithms are then applied to
modify the topology of the tree and increase its likelihood until
convergence on the maximum-likelihood tree is achieved.
Inference of a phylogenetic tree for the purpose of APR should
be considered an iterative process; problematic sequences should
be identified and removed, and alternative sequences should be
added before repeating the phylogenetic analysis. The final tree
should be biologically plausible, supported by high bootstrap values, and robust to variations in the evolutionary model used.
1. Open the PhyML [25] web interface and upload the sequence
alignment.
2. Select “Amino-Acids” for data type.
3. Under “Substitution Model,” specify the substitution model
chosen in Subheading 3.3. +I denotes an estimated proportion of invariant sites; +G denotes an estimated gamma-shape
parameter, and +F denotes empirical amino acid frequencies.
4. Under “Tree Searching,” select “BIONJ” for the starting tree and
“NNI + SPR” for the type of tree improvement (see Note 10).
5. Under “Branch Support,” select the “aLRT SH-like” branch
support test. Alternatively, for the final tree, select a bootstrap
calculation with 100 replicates.
6. Once the analysis has completed, open the maximum-likelihood tree in FigTree. The tree can be rooted by selecting the
branch connecting the ingroup and outgroup sequences and
clicking the “Reroot” button.
7. Inspect the tree to identify problems such as long branches
and low branch support values, and rectify these problems by
repeating the multiple sequence alignment and phylogenetic
analysis using a modified sequence dataset if necessary
(see Notes 11 and 12).
3.4 Phylogenetic
Tree Inference
Improving FRET Sensors by Ancestral Gene Resurrection