Phylogenetics pt 2

Molecular Phylogenetics

Introduction to Molecular Phylogenetics

• In a population, random mutations occur in DNA or protein sequences over time. These accumulated mutations serve as molecular markers, allowing us to trace the evolutionary history and relationships (phylogenies) among organisms. The more similar the mutation patterns, the closer the evolutionary relationship.

• The specific nucleotide (A, T, C, G) or amino acid identity at a given position (locus) in a sequence becomes the 'character state' used for phylogenetic analysis. By comparing these character states across different species or individuals, we can infer their ancestry.

Molecular Phylogenetic Methodology

1. Choose gene(s) of interest- This step is critical for successful phylogenetic reconstruction. Genes chosen should ideally be single-copy (to avoid paralogy issues), evolve at an appropriate rate (not too fast to cause saturation, not too slow to lack informative variation), and be present in all taxa under study. Examples include ribosomal RNA genes (16S, 18S) for deep phylogenies, or mitochondrial genes for shallower relationships.

2. Identify homologs- Homologous sequences, which share a common evolutionary origin, must be located for analysis. This typically involves searching genetic databases (e.g., GenBank) using tools like BLAST. It's important to distinguish between orthologs (genes diverged by a speciation event) and paralogs (genes diverged by a gene duplication event within a genome), as orthologs are usually preferred for species tree reconstruction.

3. Align sequences- Once homologous sequences are identified, they must be aligned. Alignment is essential to establish positional homology, meaning that corresponding nucleotides or amino acids across different sequences are indeed descended from a common ancestral residue. Gaps are introduced to account for insertions or deletions (indels) that have occurred over evolutionary time. Software like ClustalW or MAFFT are commonly used, and the resulting alignment represents a critical hypothesis of evolutionary correspondence.

4. Calculate gene tree- After obtaining a high-quality alignment, a phylogenetic tree is generated based on the aligned sequences. This involves applying specific algorithms (like Neighbor Joining, Maximum Likelihood, or Maximum Parsimony) to infer the branching pattern and evolutionary distances, producing a phylogenetic hypothesis.

Homology in Molecular Phylogenetics

• Homology: In molecular phylogenetics, homology is primarily established through sequence alignment. If two sequences align well and show significant similarity, it suggests they descended from a common ancestor. For example, aligning DNA sequences from various Operational Taxonomic Units (OTUs), which can be individual organisms, populations, or species, helps determine their evolutionary relationships by identifying shared ancestral sequence features.

• Sequence Alignment:

  • Every alignment is a hypothesis representing the evolutionary relationships and events (like mutations, insertions, and deletions) that have occurred in DNA or protein sequences. It postulates which residues in different sequences are evolutionarily equivalent.

  • Each position (column) in the optimal alignment functions as a character state. For example, if at a specific column, all species have an 'A', it suggests a conserved ancestral state. If some have 'A' and others 'G', it indicates a substitution event.

  • The locus used for analysis must exhibit sufficient variation among OTUs to be phylogenetically informative, meaning there are enough differences to resolve relationships, but not so much that the signal is lost due to saturation. Additionally, it must generally be a single-copy gene to avoid confounding issues from paralogous sequences, which could lead to an inaccurate representation of the species' evolutionary history.

Approaches to Molecular Phylogenetic Reconstruction

1. Neighbor Joining:- A fast, distance-based phenetic method that uses a distance matrix derived from pairwise comparisons of an alignment to cluster less distant sequences together. The algorithm iteratively finds pairs of operational taxonomic units (OTUs) that minimize the total branch length at each step. These selected OTUs are then treated as a single OTU, and a new distance matrix is computed. It's particularly useful for large datasets.

   • Example distance matrix (derived from similarities or disimilarities, e.g., number of substitutions per site):

     • Human: 26, 9, 8, 13

     • Baboon: 7, 10, 7, 10, 13

     • Cow: 3, 11, 12, 16

     • Sheep: 12, 9, 15

     • Mouse: 7, 16

     • Hamster: 14

     • Chicken

2. Maximum Likelihood:- A statistical, model-based method that evaluates different tree topologies and branch lengths to determine the tree that maximizes the likelihood of observing the given aligned sequence data. It uses explicit evolutionary models to predict the probability of character state changes along each branch.

   • The likelihood function is denoted as: $L = P(D|H)$, where D represents the observed sequence data, and H is the phylogenetic hypothesis (the specific tree topology, branch lengths, and model parameters). The goal is to find the tree H that makes the observed data D most probable.

   • For even a small number of taxa, the number of possible tree topologies can be immense. For 4 taxa, there are 3 possible unrooted tree topologies. If we consider 2 unknown internal nodes, and each node can be one of the 4 possible bases (A, T, C, G), there could be $4^2 = 16$ unique possible states at these two ancestral nodes that need to be evaluated under a given tree topology, leading to complex calculations, especially when considering different possible placements of mutations and their probabilities.

3. Maximum Parsimony:- An algorithm that seeks the simplest tree, requiring the fewest total evolutionary changes (e.g., nucleotide substitutions or morphological character state changes) to explain the observed sequence data. It operates on the principle of Occam's Razor, favoring the hypothesis that minimizes character state transformations.

   • Example: When choosing between different tree topologies for a set of species, maximum parsimony would select the tree that exhibits the minimum number of character changes (e.g., mutations from A to G, or the gain/loss of a trait) across all branches to reconcile the observed character states at the tips.

Mutation Probabilities

• Substitution models: These mathematical models describe the probabilities of one nucleotide mutating into another over a given time frame. They are crucial for distance-based and likelihood-based phylogenetic methods.

  • The simplest substitution model is the Jukes-Cantor (JC69) model, which assumes all nucleotide substitutions (A \ G, A \ C, etc.) are equally probable, designated as α. In this model, the base frequencies are also assumed to be equal (0.25 for each base).

  • However, reality shows that transitions (purine to purine, A \ G or pyrimidine to pyrimidine, C \ T mutations) are generally more common than transversions (purine to pyrimidine or vice versa, e.g., A \ C, G \ T). This biological reality is often due to the chemical properties of bases and repair mechanisms.

  • This results in a need for more complex models, such as the Kimura 2-parameter (K2P) model, which incorporates differing rates for transitions and transversions. Even more complex models, like the General Reversible Model (GTR), allow for all six possible substitution rates to be different and often account for unequal base frequencies to enhance accuracy in phylogenetic analysis.

Maximum Parsimony Example

• Observations of various vertebrates with respect to derived traits (apomorphies) showcase the utility of maximum parsimony in constructing phylogenetic trees based on shared traits. The goal is to minimize the number of independent origins or losses of these traits on the tree.

• Example list of derived traits across taxa:

  • Hagfish: - (none, typically used as an outgroup)

  • Perch: + (jaws)

  • Salamander: + + (jaws, lungs)

  • Lizard: + + + (jaws, lungs, claws)

  • Crocodile: + + + (jaws, lungs, claws)

  • Pigeon: + + + + (jaws, lungs, claws, feathers/wings, though often simplified to just 'feathers')

  • Mouse: + + + + + (jaws, lungs, claws, mammary glands, fur)

  • Chimpanzee: + + + + + + + (jaws, lungs, claws, mammary glands, fur, opposable thumb, large brain)

• Analysis:- The presence (+) or absence (-) of these derived traits are used to group species according to shared characteristics. By considering the outgroup (e.g., Hagfish) which presumably lacks all derived traits ancestral to the ingroup, we can establish the ancestral state for each character. Then, we reconstruct the tree by adding evolutionary changes (gains of traits) at the most parsimonious (fewest changes) positions on the tree. For instance, the acquisition of 'jaws' defines a major lineage, followed by 'lungs' distinguishing amphibians and subsequent groups, and so on.

Pitfalls in Tree Construction

• “Too much time” pitfall:- As the evolutionary time since lineages split increases, the probability of multiple substitutions occurring at the same site, or reversions to the ancestral character state, rises significantly. This phenomenon, known as saturation, can obscure the true phylogenetic signal.

  • When saturation occurs, distantly related taxa might appear more closely related than they actually are due to convergent evolution (e.g., both evolving to the same nucleotide by chance) or reversions. This can lead to a phenomenon known as long branch attraction (LBA), where rapidly evolving or distantly related taxa are erroneously grouped together, artifactually forming a clade because their high number of accumulated, independent changes makes them appear similar.

• “Too short time” pitfall:- When speciation events occur rapidly or lineages diverge recently, there may not have been enough time for distinct lineages to sort out their ancestral genetic polymorphisms completely. This can lead to incomplete lineage sorting (ILS), where ancestral polymorphisms are retained through speciation events, resulting in individual gene trees that do not perfectly reflect the species tree topology.

  • This means that different gene trees (phylogenies constructed from different genomic loci) for the same set of species may yield conflicting evolutionary interpretations. This discordance can be a significant challenge in resolving rapid radiations or recent divergences, making it difficult to infer the true species phylogeny from a single gene.

Reticulate Evolution

• Reticulate evolution refers to evolutionary events that involve the merging or exchange of genetic material between distinct lineages, thus forming a net-like or reticulate pattern rather than a simple, bifurcating tree structure. Key examples include:

  • Horizontal gene transfer (HGT): The non-sexual movement of genetic material between organisms, common in prokaryotes. HGT can rapidly introduce new traits and complicate phylogenetic reconstruction, as different genes might have different evolutionary histories.

  • Hybridization: The interbreeding of individuals from genetically distinct populations or species, resulting in offspring with mixed ancestry. This is particularly common in plants but also occurs in animals, including the well-known example of big cat hybrids.

  • Endosymbiosis: A specific type of reticulate evolution where one organism lives inside another, ultimately leading to a permanent genetic and functional integration. The origin of mitochondria and chloroplasts in eukaryotic cells through the engulfment of bacterial ancestors is a classic example, where genes from the endosymbiont were transferred to the host nucleus.

• An illustrative example of reticulate evolution in humans includes traces of Neandertal and Denisovan DNA retained within the genomes of modern non-African human populations, indicating past interbreeding events between distinct hominin lineages.