Prospects for building the tree of life from large sequence databases

Area of Science:

• Bioinformatics
• Molecular Evolution
• Computational Biology

Background:

• Large-scale protein sequence databases like Swiss-Prot and GenBank are crucial resources for biological research.
• Assessing the phylogenetic utility of these vast datasets is essential for understanding evolutionary history.
• Database sampling biases can introduce challenges in constructing informative phylogenetic datasets.

Purpose of the Study:

• To evaluate the phylogenetic potential of protein sequences from major biological databases.
• To determine if informative subsets can be extracted despite database limitations.
• To assess the feasibility of phylogenetic analysis with highly incomplete datasets.

Main Methods:

• Analysis of approximately 300,000 protein sequences from Swiss-Prot and GenBank.
• Identification of phylogenetically informative subsets within the larger dataset.
• Construction and analysis of "supermatrices" with substantial amounts of missing data (up to 92%).

Main Results:

• A small subset of protein sequences retained significant taxonomic diversity.
• Phylogenetic datasets derived from these databases often contain a high proportion of missing entries due to sampling biases.
• Analysis of supermatrices demonstrated that even with up to 92% missing data, broad evolutionary insights are achievable.

Conclusions:

• Protein sequence databases hold phylogenetic potential, even when data is sparse.
• Phylogenetic analyses are feasible with highly incomplete datasets, offering insights into the tree of life.
• Careful data selection and analysis methods can overcome database limitations for evolutionary studies.