A priori assessment of data quality in molecular phylogenetics

The Free Lunch is not over yet-systematic exploration of numerical thresholds in maximum likelihood phylogenetic inference

Summary

Maximum likelihood (ML) is a widely used phylogenetic inference method. ML implementations heavily rely on numerical optimization routines that use internal numerical thresholds to determine convergence. We systematically analyze the impact of these threshold settings on the log-likelihood and runtimes for ML tree inferences with RAxML-NG, IQ-TREE, and FastTree on empirical datasets. We provide empirical evidence that we can substantially accelerate tree inferences with RAxML-NG and IQ-TREE by changing the default values of two such numerical thresholds. At the same time, altering these settings does not significantly impact the quality of the inferred trees. We further show that increasing both thresholds accelerates the RAxML-NG bootstrap without influencing the resulting support values. For RAxML-NG, increasing the likelihood thresholds ϵ LnL and ϵ brlen to 10 and 103, respectively, results in an average tree inference speedup of 1.9 ± 0.6 on Data collection 1, 1.8 ± 1.1 on Data collection 2, and 1.9 ± 0.8 on Data collection 2 for the RAxML-NG bootstrap compared to the runtime under the current default setting. Increasing the likelihood threshold ϵ LnL to 10 in IQ-TREE results in an average tree inference speedup of 1.3 ± 0.4 on Data collection 1 and 1.3 ± 0.9 on Data collection 2.

Availability and implementation

All MSAs we used for our analyses, as well as all results, are available for download at https://cme.h-its.org/exelixis/material/freeLunch_data.tar.gz. Our data generation scripts are available at https://github.com/tschuelia/ml-numerical-analysis.

From Easy to Hopeless-Predicting the Difficulty of Phylogenetic Analyses

Phylogenetic analyzes under the Maximum-Likelihood (ML) model are time and resource intensive. To adequately capture the vastness of tree space, one needs to infer multiple independent trees. On some datasets, multiple tree inferences converge to similar tree topologies, on others to multiple, topologically highly distinct yet statistically indistinguishable topologies. At present, no method exists to quantify and predict this behavior. We introduce a method to quantify the degree of difficulty for analyzing a dataset and present Pythia, a Random Forest Regressor that accurately predicts this difficulty. Pythia predicts the degree of difficulty of analyzing a dataset prior to initiating ML-based tree inferences. Pythia can be used to increase user awareness with respect to the amount of signal and uncertainty to be expected in phylogenetic analyzes, and hence inform an appropriate (post-)analysis setup. Further, it can be used to select appropriate search algorithms for easy-, intermediate-, and hard-to-analyze datasets.

Phylogeny of Hawaiian Melicope (Rutaceae): RAD-seq Resolves Species Relationships and Reveals Ancient Introgression

Hawaiian Melicope are one of the major adaptive radiations of the Hawaiian Islands comprising 54 endemic species. The lineage is monophyletic with an estimated crown age predating the rise of the current high islands. Phylogenetic inference based on Sanger sequencing has not been sufficient to resolve species or deeper level relationships. Here, we apply restriction site-associated DNA sequencing (RAD-seq) to the lineage to infer phylogenetic relationships. We employ Quartet Sampling to assess information content and statistical support, and to quantify discordance as well as partitioned ABBA-BABA tests to uncover evidence of introgression. Our new results drastically improved resolution of relationships within Hawaiian Melicope. The lineage is divided into five fully supported main clades, two of which correspond to morphologically circumscribed infrageneric groups. We provide evidence for both ancestral and current hybridization events. We confirm the necessity for a taxonomic revision of the Melicope section Pelea, as well as a re-evaluation of several species complexes by combining genomic and morphological data.

Biodiversity seen through the perspective of insects: 10 simple rules on methodological choices and experimental design for genomic studies

Massively parallel DNA sequencing opens up opportunities for bridging multiple temporal and spatial dimensions in biodiversity research, thanks to its efficiency to recover millions of nucleotide polymorphisms. Here, we identify the current status, discuss the main challenges, and look into future perspectives on biodiversity genomics focusing on insects, which arguably constitute the most diverse and ecologically important group among all animals. We suggest 10 simple rules that provide a succinct step-by-step guide and best-practices to anyone interested in biodiversity research through the study of insect genomics. To this end, we review relevant literature on biodiversity and evolutionary research in the field of entomology. Our compilation is targeted at researchers and students who may not yet be specialists in entomology or molecular biology. We foresee that the genomic revolution and its application to the study of non-model insect lineages will represent a major leap to our understanding of insect diversity.

Topological support and data quality can only be assessed through multiple tests in reviewing Blattodea phylogeny

Assessing support for molecular phylogenies is difficult because the data is heterogeneous in quality and overwhelming in quantity. Traditionally, node support values (bootstrap frequency, Bayesian posterior probability) are used to assess confidence in tree topologies. Other analyses to assess the quality of phylogenetic data (e.g. Lento plots, saturation plots, trait consistency) and the resulting phylogenetic trees (e.g. internode certainty, parameter permutation tests, topological tests) exist but are rarely applied. Here we argue that a single qualitative analysis is insufficient to assess support of a phylogenetic hypothesis and relate data quality to tree quality. We use six molecular markers to infer the phylogeny of Blattodea and apply various tests to assess relationship support, locus quality, and the relationship between the two. We use internode-certainty calculations in conjunction with bootstrap scores, alignment permutations, and an approximately unbiased (AU) test to assess if the molecular data unambiguously support the phylogenetic relationships found. Our results show higher support for the position of Lamproblattidae, high support for the termite phylogeny, and low support for the position of Anaplectidae, Corydioidea and phylogeny of Blaberoidea. We use Lento plots in conjunction with mutation-saturation plots, calculations of locus homoplasy to assess locus quality, identify long branch attraction, and decide if the tree's relationships are the result of data biases. We conclude that multiple tests and metrics need to be taken into account to assess tree support and data robustness.

Plastome Rearrangements in the "Adenocalymma-Neojobertia" Clade (Bignonieae, Bignoniaceae) and Its Phylogenetic Implications

The chloroplast is one of the most important organelles of plants. This organelle has a circular DNA with approximately 130 genes. The use of plastid genomic data in phylogenetic and evolutionary studies became possible with high-throughput sequencing methods, which allowed us to rapidly obtain complete genomes at a reasonable cost. Here, we use high-throughput sequencing to study the "Adenocalymma-Neojobertia" clade (Bignonieae, Bignoniaceae). More specifically, we use Hi-Seq Illumina technology to sequence 10 complete plastid genomes. Plastomes were assembled using selected plastid reads and de novo approach with SPAdes. The 10 assembled genomes were analyzed in a phylogenetic context using five different partition schemes: (1) 91 protein-coding genes ("coding"); (2) 76 introns and spacers with alignment manually edited ("non-coding edited"); (3) 76 non-coding regions with poorly aligned regions removed using T-Coffee ("non-coding filtered"); (4) 91 coding regions plus 76 non-coding regions edited ("coding + non-coding edited"); and, (5) 91 protein-coding regions plus the 76 filtered non-coding regions ("coding + non-coding filtered"). Fragmented regions were aligned using Mafft. Phylogenetic analyses were conducted using Maximum Likelihood (ML) and Bayesian Criteria (BC). The analyses of the individual plastomes consistently recovered an expansion of the Inverted Repeated (IRs) regions and a compression of the Small Single Copy (SSC) region. Major genomic translocations were observed at the Large Single Copy (LSC) and IRs. ML phylogenetic analyses of the individual datasets led to the same topology, with the exception of the analysis of the "non-coding filtered" dataset. Overall, relationships were strongly supported, with the highest support values obtained through the analysis of the "coding + non-coding edited" dataset. Four regions at the LSC, SSC, and IR were selected for primer development. The "Adenocalymma-Neojobertia" clade shows an unusual pattern of plastid structure variation, including four major genomic translocations. These rearrangements challenge the current view of conserved plastid genome architecture in terms of gene order. It also complicates both genomic assemblies using reference genomes and sequence alignments using whole plastomes. Therefore, strategies that employ de novo assemblies and manual evaluation of sequence alignments are required to prevent assembly and alignment errors.

Power, resolution and bias: recent advances in insect phylogeny driven by the genomic revolution

Our understanding on the phylogenetic relationships of insects has been revolutionised in the last decade by the proliferation of next generation sequencing technologies (NGS). NGS has allowed insect systematists to assemble very large molecular datasets that include both model and non-model organisms. Such datasets often include a large proportion of the total number of protein coding sequences available for phylogenetic comparison. We review some early entomological phylogenomic studies that employ a range of different data sampling protocols and analyses strategies, illustrating a fundamental renaissance in our understanding of insect evolution all driven by the genomic revolution. The analysis of phylogenomic datasets is challenging because of their size and complexity, and it is obvious that the increasing size alone does not ensure that phylogenetic signal overcomes systematic biases in the data. Biases can be due to various factors such as the method of data generation and assembly, or intrinsic biological feature of the data per se, such as similarities due to saturation or compositional heterogeneity. Such biases often cause violations in the underlying assumptions of phylogenetic models. We review some of the bioinformatics tools available and being developed to detect and minimise systematic biases in phylogenomic datasets. Phylogenomic-scale data coupled with sophisticated analyses will revolutionise our understanding of insect functional genomics. This will illuminate the relationship between the vast range of insect phenotypic diversity and underlying genetic diversity. In combination with rapidly developing methods to estimate divergence times, these analyses will also provide a compelling view of the rates and patterns of lineagenesis (birth of lineages) over the half billion years of insect evolution.