Publications by authors named "Ben Bettisworth"
Article Synopsis
- Researchers used methods from molecular evolution to calculate exact win probabilities for knockout tournaments instead of using slower simulations.
- Their new open-source code is significantly faster, offering improvements two orders of magnitude faster than simulating and over two times faster than traditional methods.
- They can now quickly assess prediction uncertainty by running extensive analyses, even for larger tournaments, on standard laptops in just minutes.
Article Synopsis
- The DEC model of biogeography is computationally intensive, requiring significant processing time for large datasets, limiting analyses based on the number of regions.
- The newly developed tool, Lagrange-NG, offers up to 49 times faster performance with multithreading and 26 times faster with a single thread, enabling efficient analysis of up to 12 regions in about 18 minutes.
- Lagrange-NG not only improves computational speed by using Krylov subspaces for matrix exponential calculations, but also adheres to higher coding quality standards, making it a reliable and accessible tool for researchers under GPL2.
Article Synopsis
- Phylogenetic analysis using the Maximum-Likelihood model is complex and requires significant time and resources, as it involves inferring multiple independent trees from datasets.
- Depending on the dataset, results can show consistent tree structures or vastly different topologies that are statistically similar, but current methods can't predict which will happen.
- The new tool Pythia, a Random Forest Regressor, helps quantify the difficulty of analyzing datasets in terms of expected signal and uncertainty, allowing researchers to choose suitable analysis setups and algorithms based on the dataset's complexity.
Article Synopsis
- Recent advancements in full-genome sequencing have led to phylogenetic analyses that involve extremely long sequences, making their computational analysis challenging and often requiring powerful clusters due to high resource demands.* -
- This study introduces an AI-driven method using Lasso-regression, allowing researchers to efficiently select a small, optimal subset of sites (as little as 5%) that significantly simplifies the analysis while still accurately estimating the tree structure.* -
- The proposed code is available on GitHub, allowing for easy access and implementation, and it has demonstrated reduced computational time without sacrificing accuracy in phylogenetic tree search performance.*
Background: In phylogenetic analysis, it is common to infer unrooted trees. However, knowing the root location is desirable for downstream analyses and interpretation. There exist several methods to recover a root, such as molecular clock analysis (including midpoint rooting) or rooting the tree using an outgroup.
View Article and Find Full Text PDFArticle Synopsis
- Many daily publications focus on analyzing SARS-CoV-2 data, including phylogenetic studies available on nextstrain.org.
- The authors discuss challenges in creating reliable phylogenies due to a high number of virus sequences but a low number of mutations, making it tough to draw clear evolutionary connections.
- They conclude that while phylogenetic methods can offer some insights into COVID-19's evolution and spread, researchers should interpret results with caution, especially when using standard analysis tools.