FitMultiModel#
HyPhy version required ≥ 2.5.7
Parallel support MP
File path TemplateBatchFiles/SelectionAnalyses/FitMultiModel.bf
Standard analysis menu FMM
Citation Preprint

What biological question is the method designed to answer?#

Does a data set have evidence of multiple simultaneous hits (MH) and to what exitent?

FMM can be useful if a scientist wishes to see if MH exsist in their data and the potential impact of those MH. FMM will also provide a list of sites that show strong preference for codon models with multiple hits that may be used for further exploration.

What is the statistical procedure and statistical test is used to establish significance for this method?#

For each alignment the following models are fit to the data:

  • 1H: The standard Muse and Gaut (1994) codon model that allows for only single instantaneous nucleotide changes

  • 2H: an extension of the 1H model that allows two nucleotides in a codon to be substituted instantaneously with rate δ relative to 1H synonymous rate

  • 3HSI: the 2H model extended to allow three simultaneous nucleotide substitutions in a codon if the change is synonymous with relative rate, ψs

  • 3H: the 2H model extend to allow any three nucleotide simultanceous substitutions with relative rate ψ

  • 3H+: the 3HSI model extended to permit any three-nulceotide substitutions with relative rate ψ

See the following table for a breakdown of parameters and if they are estimated for each model:

Model
Parameter Description 1H 2H 3HSI 3H 3H+
ωi Site dN/dS ratio Random effect 3-bin GDD distribution
δ Global 2H/1H rate ratio 0 Estimated Estimated Estimated Estimated
ψs Global 3H/1H rate ratio for synonymous codon islands 0 Estimated Estimated = ψ Estimated
ψ Global 3H/1H rate ratio 0 0 0 Estimated Estimated

The actual statistical test being performed are pairwise likelihood ratio test between the models such that the null is always the model with fewer parameters eg: 1H v 2H, 1H - null and 2H alternative.

How should one interpret positive and negative test results?#

A positive result (ie p ≤ 0.05) for a likelihood ratio test between two models, H0 and HA, means that we reject H0. For example, if we look at a LRT between 1H and 2H, where 1H is H0 and 2H is HA, a p value ≤ 0.05 means we can reject the hypothesis that 2H has zero rates.

A negative result (ie p > 0.05) means we fail to reject the null. Thus in the same example as above (1H vs 2H) we would fail to reject the hypothesis that 2H has zero rates.

Rules of thumb for when this method is likely to work well, and when it is not.#

This method is likely to work well on trees with longer branches and sequences. It is not likely to work well on shorter trees.

Example#

For more information about running options see here

HYPHYMP FitMultiModel.bf --alignment p51.nex --triple-islands Yes

The following data are output to the screen.

Analysis Description#

Examine whether or not a codon alignment is better fit by models which permit multiple instantaneous substitutions. v0.2 adds a separate rate for codon-island triple-hit rates

  • Requirements: in-frame codon alignment and a phylogenetic tree

  • Written by: Sergei L Kosakovsky Pond, Sadie Wisotsky and Alexander Lucaci

  • Contact Information: spond@temple.edu

  • Analysis Version: 0.2

code –> Universal Loaded a multiple sequence alignment with 8 sequences, 440 codons, and 1 partitions from /home/swisotsky/hyphy-analyses/FitMultiModel/p51.nex The number of omega rate classes to include in the model (permissible range = [1,10], default value = 3, integer): rates –> 3

triple-islands –> Yes

Obtaining branch lengths and nucleotide substitution biases under the nucleotide GTR model#

  • Log(L) = -3320.50, AIC-c = 6683.09 (21 estimated parameters)

Obtaining the global omega estimate based on relative GTR branch lengths and nucleotide substitution biases#

  • Log(L) = -3178.60, AIC-c = 6413.66 (28 estimated parameters)
  • non-synonymous/synonymous rate ratio for test = 0.2459

Fitting Standard MG94#

  • Log(L) = -3121.04, AIC-c = 6308.73 (33 estimated parameters)
  • non-synonymous/synonymous rate ratio = 0.2967
  • The following relative rate distribution (mean 1) for site-to-site non-synonymous rate variation was inferred
Rate Proportion, % Notes
0.186 86.040
4.466 13.674
80.143 0.286

Fitting MG94 with double instantaneous substitutions#

  • Log(L) = -3121.03, AIC-c = 6310.74 (34 estimated parameters)
  • non-synonymous/synonymous rate ratio = 0.2946
  • rate at which 2 nucleotides are changed instantly within a single codon = 0.0127
  • The following relative rate distribution (mean 1) for site-to-site non-synonymous rate variation was inferred
Rate Proportion, % Notes
0.186 85.954
4.448 13.757
79.054 0.289

Fitting MG94 with double and triple instantaneous substitutions#

  • Log(L) = -3121.03, AIC-c = 6314.82 (36 estimated parameters)
  • non-synonymous/synonymous rate ratio = 0.2947
  • rate at which 2 nucleotides are changed instantly within a single codon = 0.0127
  • rate at which 3 nucleotides are changed instantly within a single codon = 0.0000
  • rate at which 3 nucleotides are changed instantly within a single codon between synonymous codon islands = 0.0000
  • The following relative rate distribution (mean 1) for site-to-site non-synonymous rate variation was inferred
Rate Proportion, % Notes
0.186 85.952
4.444 13.756
78.643 0.291

Fitting MG94 with double and triple instantaneous substitutions [only synonymous islands]#

  • Log(L) = -3121.03, AIC-c = 6312.78 (35 estimated parameters)
  • non-synonymous/synonymous rate ratio = 0.2946
  • rate at which 2 nucleotides are changed instantly within a single codon = 0.0126
  • rate at which 3 nucleotides are changed instantly within a single codon = 0.0000
  • rate at which 3 nucleotides are changed instantly within a single codon between synonymous codon islands = 0.0000
  • The following relative rate distribution (mean 1) for site-to-site non-synonymous rate variation was inferred
Rate Proportion, % Notes
0.186 85.959
4.451 13.751
78.760 0.290

Summary of rate estimates and significance testing#

Model Log-L omega 2-hit rate p-value 3-hit rate p-value
Standard MG94 -3121.04 0.2967 N/A N/A N/A N/A
Standard MG94 + 2 hits -3121.03 0.2946 0.0127 0.8622 (2-hit rate = 0) N/A N/A
Standard MG94 + 3 hits (islands) -3121.03 0.2946 0.0000 0.9960 (3-hit island vs 2-hit) 0.0000 1.0000 (3-hit = 0)
Standard MG94 + 2 or 3 hits -3121.03 0.2947 0.0127 0.9986 (2&3-hit rates = 0) 0.0000 1.0000 (3-hit rate(s) = 0)

No individual sites showed sufficiently strong preference for multiple-hit models#

Writing detailed analysis report to `/home/swisotsky/hyphy-analyses/FitMultiModel/p51.nex.FITTER.json'#