Fig 1.
Graphical representation of the binning effect of using alignment-ordered versus staggered sequence input order for “top-down” centroid-based clustering. Shaded rectangles represent sequences, where the shade consistently portrays a specific sequence throughout the diagram. The multiple alignment on the left shows each of the sequences ordered based on fractional identity, where nearby sequences are more closely related than distant ones, and distributed evenly across a fractional identity range of 0.1. For both aligned and staggered input ordering, sequences are read from top to bottom by the UCLUST algorithm of USEARCH and either placed in a cluster that has the best match to the centroid sequence above the given identity cutoff, or is made the centroid sequence of a new cluster if a match cannot be found. In this diagram, centroid sequences are the top sequences of each cluster. With the aligned input order, it is shown that some sequences can be binned in clusters that do not contain their closest centroid match. The staggered input places sequences in correct bins essentially by first defining all centroid sequences.
Fig 2.
Dendritic heat map representing the fifth mutation step of the simulated mutation lineage data generated as described in the methods and clustered using the average-linkage algorithm of the “bottom-up” method. The darkly colored wedge at the 0° position represents the minimum (red) and maximum (blue) possible heat map relative abundance bin responses, GC≤50% and GC>50% respectively. White space in the heat maps represents clusters with neutral bin response. Rings, starting at the center, represent clusters of sequences for identity cutoffs of 0.75 to 1.0. Clusters, including single-sequence clusters, are plotted in a radial range that is conserved from the clusters from which they were derived. High resolution versions of all DHMs in this manuscript are available in S2 and S3 Files.
Fig 3.
Dendritic heat maps from bottom-up minimum-linkage hierarchical clustering of a mutating population.
Dendritic heat maps representing 0 through 15 mutations of the simulated mutation lineage data generated as described in the methods and clustered using the minimum-linkage algorithm of the “bottom-up” method. Panel zero represents the most homologous set of sequences (identical) and panel fifteen represents the least homologous set of sequences (fifteen base substitutions). The darkly colored wedge at the 0° position of each dendritic heat map represents the minimum (red) and maximum (blue) possible heat map relative abundance bin responses of all dendritic heat maps displayed, GC≤50% and GC>50% respectively. White space in the heat maps represents clusters with neutral bin response. Rings, starting at the center, represent clusters of sequences for identity cutoffs of 0.75 to 1.0. Clusters, including single-sequence clusters, are plotted in a radial range that is conserved from the clusters from which they were derived. High resolution versions of all DHMs in this manuscript are available in S2 and S3 Files.
Fig 4.
Dendritic heat maps from bottom-up maximum-linkage hierarchical clustering of a mutating population.
Dendritic heat maps representing 0 through 15 mutations of the simulated mutation lineage data generated as described in the methods and clustered using the maximum-linkage algorithm of the “bottom-up” method. Panel zero represents the most homologous set of sequences (identical) and panel fifteen represents the least homologous set of sequences (fifteen base substitutions). The darkly colored wedge at the 0° position of each dendritic heat map represents the minimum (red) and maximum (blue) possible heat map relative abundance bin responses of all dendritic heat maps displayed, GC≤50% and GC>50% respectively. White space in the heat maps represents clusters with neutral bin response. Rings, starting at the center, represent clusters of sequences for identity cutoffs of 0.75 to 1.0. Clusters, including single-sequence clusters, are plotted in a radial range that is conserved from the clusters from which they were derived. High resolution versions of all DHMs in this manuscript are available in S2 and S3 Files.
Fig 5.
Dendritic heat maps from bottom-up average-linkage hierarchical clustering of a mutating population.
Dendritic heat maps representing 0 through 15 mutations of the simulated mutation lineage data generated as described in the methods and clustered using the average-linkage algorithm of the “bottom-up” method. Panel zero represents the most homologous set of sequences (identical) and panel fifteen represents the least homologous set of sequences (fifteen base substitutions). The darkly colored wedge at the 0° position of each dendritic heat map represents the minimum (red) and maximum (blue) possible heat map relative abundance bin responses of all dendritic heat maps displayed, GC≤50% and GC>50% respectively. White space in the heat maps represents clusters with neutral bin response. Rings, starting at the center, represent clusters of sequences for identity cutoffs of 0.75 to 1.0. Clusters, including single-sequence clusters, are plotted in a radial range that is conserved from the clusters from which they were derived. High resolution versions of all DHMs in this manuscript are available in S2 and S3 Files.
Fig 6.
Dendritic heat maps from top-down hierarchical clustering of a mutating population.
Dendritic heat maps representing 0 through 15 mutations of the simulated mutation lineage data generated as described in the methods and clustered using the “top-down” method. Panel zero represents the most homologous set of sequences (identical) and panel fifteen represents the least homologous set of sequences (fifteen base substitutions). The darkly colored wedge at the 0° position of each dendritic heat map represents the minimum (red) and maximum (blue) possible heat map relative abundance bin responses of all dendritic heat maps displayed, GC≤50% and GC>50% respectively. White space in the heat maps represents clusters with neutral bin response. Rings, starting at the center, represent clusters of sequences for identity cutoffs of 0.75 to 1.0. Clusters, including single-sequence clusters, are plotted in a radial range that is conserved from the clusters from which they were derived. High resolution versions of all DHMs in this manuscript are available in S2 and S3 Files.
Fig 7.
Dendritic heat maps from top-down hierarchical clustering of a growing population.
Dendritic heat maps representing generations 0 through 15 of the simulated population growth data generated as described in the methods and clustered using the “top-down” method. The darkly colored wedge at the 0° position of each dendritic heat map represents the minimum (red) and maximum (blue) possible heat map relative abundance bin responses of all dendritic heat maps displayed, GC ≤50% and GC>50% respectively. White space represents clusters with neutral bin response. Rings, starting at the center, represent clusters of sequences for identity cutoffs of 0.75 to 1.0. Clusters, including single-sequence clusters, are plotted in a radial range that is conserved from the clusters from which they were derived. High resolution versions of all DHMs in this manuscript are available in S2 and S3 Files.
Fig 8.
Fig 3 reused from Elser et al. 2014 with the kind permission of ASM.
Dendritic heat maps displaying habitat preferences for multiple levels of phylogenetic clades across multiple time points and locations. Reprinted from [33] under a CC BY license, with permission from AEM, original copyright 2014.
Fig 9.
Fig 4 reused from Elser et al. 2014 with the kind permission of ASM.
Histograms displaying the strongest habitat preferences for the phylum, order, and genus taxonomical levels of four sample types. A skew line is used to show the relative strength of habitat preference. Reprinted from [33] under a CC BY license, with permission from AEM, original copyright 2014.
Fig 10.
Fig 1 reused from Eisen et al. 1998 with the kind permission of PNAS.
Heat map displaying data from a time course of serum stimulation of primary human fibroblasts. Reprinted from [34] under a CC BY license, with permission from PNAS, original copyright 1998.