Re: Paralinear distance (LogDet) is a generalized correlation.
Explore the source record for details and available documents.
Models, Genetic↗
Biomedical subjects
J A Cavender
Publications and source records attributed to J A Cavender.
Necessary conditions for the method of inferring phylogeny by linear invariants.
It is known that if all the Markov transition matrices that govern the substitution of one nucleotide for another satisfy six linear constraints, then equations can be derived that permit one to infer evolutionary trees from nucleic acid sequences by the method of linear invariants. These sufficient conditions are also necessary. Any relaxation of them results in the loss of all linear invariants. Necessary conditions for any given set of linear invariants can be derived by examining conditions a matrix must satisfy to map a certain set of matrices into itself. To the extent that necessary conditions are incorrect, a method is not reliable. In a world where different parts of molecules evolve at different rates, the two-parameter model of Kimura may not be empirically distinguishable from the more general one treated here.
Base Sequence↗
Mechanized derivation of linear invariants.
Linear invariants, discovered by Lake, promise to provide a versatile way of inferring phylogenies on the basis of nucleic acid sequences (the method that he called "evolutionary parsimony"). A semigroup of Markov transition matrices embodies the assumptions underlying the method, and alternative semigroups exist. The set of all linear invariants may be derived from the semigroup by using an algorithm described here. Under assumptions no stronger than Lake's, there are greater than 50 independent linear invariants for each of the 15 rooted trees linking four species.
Algorithms↗