Rienzi.

Goodrich, Samuel G.

doi:10.1037/12002-032

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The precise meaning of clusters depends on the application. For example, when 6 analyzing 16S microbiome data, the standard pipeline is to use Operational Taxonomic 7 Units (OTUs), which are essentially clusters of closely related sequences that do not 8 diverge more than a certain threshold [1][2][3]. Another example is HIV transmission 9 inference, a field in which a dominant approach is to cluster HIV sequences from 10 different individuals based on their similarity (again using a threshold) and to use these 11 clusters as proxies to clusters of disease transmission [4,5].…”

Section: Introductionmentioning

confidence: 99%

TreeCluster: clustering biological sequences using phylogenetic trees

Balaban

Moshiri

Mai

et al. 2019

Preprint

View full text Add to dashboard Cite

Clustering homologous sequences based on their similarity is a problem that appears in many bioinformatics applications. The fact that sequences cluster is ultimately the result of their phylogenetic relationships. Despite this observation and the natural ways in which a tree can define clusters, most applications of sequence clustering do not use a phylogenetic tree and instead operate on pairwise sequence distances. Due to advances in large-scale phylogenetic inference, we argue that tree-based clustering is under-utilized. We define a family of optimization problems that, given a (not necessarily ultrametric) tree, return the minimum number of clusters such that all clusters adhere to constraints on their heterogeneity. We study three specific constraints that limit the diameter of each cluster, the sum of its branch lengths, or chains of pairwise distances. These three versions of the problem can be solved in time that increases linearly with the size of the tree, a fact that has been known by computer scientists for two of these three criteria for decades. We implement these algorithms in a tool called TreeCluster, which we test on three applications: OTU picking for microbiome data, HIV transmission clustering, and divide-and-conquer multiple sequence alignment. We show that, by using tree-based distances, TreeCluster generates more internally consistent clusters than alternatives and improves the effectiveness of downstream applications. TreeCluster is available at https://github.com/niemasd/TreeCluster. 16 which can easily scale to datasets of even millions of sequences [8]. Moreover, accurate 17 alignment of datasets with hundreds of thousands of species (a prerequisite to most 18 phylogenetic reconstruction methods) is now possible using divide-and-conquer 19 methods [9, 10]. 20 Most existing sequence clustering methods use the pairwise distances among 21 sequences as input but do not take advantage of phylogenetic trees. For example, the 22 widely-used UCLUST [2] searches for a clustering that minimizes the Hamming distance 23 of sequences to the cluster centroid while maximizes the Hamming distance between 24 centroids. Several other clustering methods have been developed for various contexts, 25 such as gene family circumscription [11, 12] and large protein sequence databases [13]. 26Using phylogenies for clustering has two potential advantages. i) Since phylogenies 27 explicitly seek to infer the evolutionary history, phylogeny-based clustering has the 28 potential to not only reflect evolutionary distances (i.e., branch lengths) but also 29 relationships (i.e., the tree topology). Recall also that branch lengths in a phylogeny are 30 model-based "corrections" of sequence distances in a statistically-rigorous way [7], and 31 therefore, may better reflect divergence between organisms. ii) When inferred using 32 subquadratic algorithms, the tree can eliminate the need to compute all pairwise 33 distances, which can improve speed and scalability. Moreover, a phylogeny often has to 34 be inferre...

show abstract

Section: Introductionmentioning

confidence: 99%