Time Trees can be computed in MEGA where divergence times are estimated for all branching points in a tree using the RelTime method (RelTime is described in Tamura et al. 2012 ) which does not require assumptions for lineage rate variations. The implementation in MEGA is very fast and expands on the RelTime method so that multiple calibration constraints can be provided, in which case MEGA will produce absolute divergence times along with relative divergence times while respecting the provided constraints. Additionally, the implementation in MEGA can compute divergence times without calibration constraints, in which case, only relative times will be produced.
There are several types of calibrations that can be used in MEGA:
Calibration densities:
Statistical densities that provide prior belief about the possible location of the true species divergence time relative to the minimum and/or maximum constraints can be used. When using this option, each calibration density is transformed into a pair of discrete constraints such that the minimum bound is placed at 2.5% of the density age and the maximum bound at the 97.5% of the density age . This means that the minimum and maximum bounds will cover 95% of the total probability density. Three statistical distribution can be used for calibration densities in MEGA:
Normal - requires that a mean and standard deviation be provided and minimum and maximum constraints will be derived from the distribution. For instance, a calibration density using a normal distribution with mean=10 and stddev=1 will produce a constraint where minTime=8.04 and maxTime=11.96
Exponential - requires that a divergence time and decay are provided and a minimum constraint will be derived from the distribution. For instance, a calibration density using an exponential distribution with mean=10 and decay=0.25 will produce a constraint where minTime=9.9
Uniform - requires that a minTime and maxTime be provided and will produce a constraint whose minTime and maxTime are those provided.
Lognormal - requires that 3 parameters are provided: offset, mean, and stddev and minimum and maximum constraints will be derived from the distribution. For instance, a calibration density using a lognormal distribution with offset=7, mean=1.5, and stddev=0.15 will produce a constraint where minTime=10.34 and maxTime=13.01
Minimum Times:
Sets a hard minimum divergence time constraint on the target node.
Maximum Times:
Sets a hard maximum divergence time constraint on the target node.
Fixed Times:
The divergence time for the target node will be equal to the provided fixed time.
Fixed Rate:
This option will define a global evolutionary rate r (in units of substitutions per site per year) that is used throughout the tree. For every node in the tree whose height (in units of substitutions per site) is h, the divergence time of the node will be set to h/r.
Tip Dates (sample times):
This option is only used for the RTDT (RelTime with Dated Tips) method. In this case, the tip dates are the dates at which molecular sequences were sampled. This method is suitable for the analysis of DNA and protein sequences from fast evolving pathogens and those generated from ancient times.
See also