BIONJ: An Improved Version of the NJ Algorithm Based on a Simple Model of Sequence Data
Citations
39,110 citations
Cites methods from "BIONJ: An Improved Version of the N..."
...…or generated automatically by applying NJ and BIONJ algorithms to a matrix of pairwise distances estimated using a maximum composite likelihood approach for nucleotide sequences and a JTT model for amino acid sequences (Saitou and Nei 1987; Jones et al. 1992; Gascuel 1997; Tamura et al. 2004)....
[...]
...The initial tree for the ML search can be supplied by the user (Newick format) or generated automatically by applying NJ and BIONJ algorithms to a matrix of pairwise distances estimated using a maximum composite likelihood approach for nucleotide sequences and a JTT model for amino acid sequences (Saitou and Nei 1987; Jones et al. 1992; Gascuel 1997; Tamura et al. 2004)....
[...]
16,261 citations
Cites methods from "BIONJ: An Improved Version of the N..."
...BIONJ is just as fast as NJ but is slightly more accurate and requires O(n3) time....
[...]
...This explains why our O(pns) ML optimization has computing time in the same range as distance methods such as NJ, BIONJ, and Weighbor, which requireO(n2s + n3) time, including distance estimation....
[...]
...An initial tree is built from this matrix, using BIONJ (Gascuel, 1997)....
[...]
14,385 citations
Cites methods from "BIONJ: An Improved Version of the N..."
...Indeed, five SPR searches with a random starting tree plus one SPR search with a BioNJ starting tree are performed for each data set, which explains the multiplicative time factor compared with SPR (one NNI search is also performed, but the computing time is negligible compared with the six SPR searches)....
[...]
...Our SPR algorithm comes in different flavors: 1) the “SPR” option relies on the algorithm described above, starting from a BioNJ (Gascuel 1997) or a maximum parsimony (MP) tree; 2) “BEST” runs both the SPR and the NNI algorithms and outputs the best of the two resulting trees (usually the SPR tree, but not always, see below); and 3) “RAND”: trees inferred by SPR starting from random trees (instead of BioNJ or MP) can also be added to the BEST option....
[...]
...Our SPR algorithm comes in different flavors: 1) the “SPR” option relies on the algorithm described above, starting from a BioNJ (Gascuel 1997) or a maximum parsimony (MP) tree; 2) “BEST” runs both the SPR and the NNI algorithms and outputs the best of the two resulting trees (usually the SPR tree,…...
[...]
...Pair agglomeration in BioNJ (used by PhyML to initiate the tree search) is an example of such choice, which should be random when some sequences are (nearly) identical....
[...]
13,668 citations
Cites methods from "BIONJ: An Improved Version of the N..."
...starts with a BIONJ tree (Gascuel 1997) whereas RAxML...
[...]
10,010 citations
Additional excerpts
...FastTree’s minimum-evolution SPR moves give it a better starting tree than PhyML’s starting tree, which is obtained with BIONJ (a weighted variant of neighbor joining [18])....
[...]
References
57,055 citations
6,956 citations
2,519 citations