# Difference between revisions of "Wilke:Using HyPhy"

Notice: The Wilke Lab page has moved to http://wilkelab.org.
The page you are looking at is kept for archival purposes and will not be further updated.
THE WILKE LAB

## General Notes

Some notes on HyPhy.

• Global variables in HYPHY must be defined before the model definition.
• How do you assign the range of the parameters in HyPhy?
myParameter :> lowerBound; /* default is 0 */

myParameter :< upperBound; /* default is 10^26 */

## Functions

/* A HYPHY translation of a Python function written by Tong Zhou to calculate the proportions of synonymous and nonsynonymous mutations, or rho. Takes as parameters a 61 x 1 frequency vector, a 61 x 61 rate matrix, and the number of species in the alignment. Returns the proportion of nonsynonymous mutations, the proportion of synonymous mutations, and the sum of all the mutations divided by the number of species, which is equal to the t parameter. Function must be placed after optimization in the script. To calculate dN or dS, the function must be called twice: once where the selection on the protein has been estimated and once where selection on the protein has been fixed.

• /

function calcRhoRate(frequency_vector, rate_matrix, species) { Codon_Index = { {14,13,14,13, 7, 7, 7, 7,19, 5,19, 5, 2, 2, 3, 2, 12,11,12,11, 6, 6, 6, 6,19,19,19,19, 1, 1, 1, 1, 16,15,16,15, 8, 8, 8, 8,20,20,20,20, 4, 4, 4, 4, 9, 9, 5, 5, 5, 5, 17,18,17, 1, 0, 1, 0} };

Codon_Index_ENG = {

{"AAA","AAC","AAG","AAT","ACA","ACC","ACG","ACT","AGA","AGC","AGG","AGT","ATA","ATC","ATG","ATT", "CAA","CAC","CAG","CAT","CCA","CCC","CCG","CCT","CGA","CGC","CGG","CGT","CTA","CTC","CTG","CTT", "GAA","GAC","GAG","GAT","GCA","GCC","GCG","GCT","GGA","GGC","GGG","GGT","GTA","GTC","GTG","GTT", "TAC", "TAT","TCA","TCC","TCG","TCT", "TGC","TGG","TGT","TTA","TTC","TTG","TTT"}

};

n = 61; sum_diag = 0.0; sum_all = 0.0; sum_syn = 0.0;

for (i=0; i<n; i=i+1) { for (j=0; j<n; j=j+1) { if (i == j) { sum_diag = sum_diag + (frequency_vector[i] * rate_matrix[i][j] * frequency_vector[j]); continue; } else { codon_i = Codon_Index[i]; codon_j = Codon_Index[j]; sum_all = sum_all + (frequency_vector[i] * rate_matrix[i][j] * frequency_vector[j]); if (codon_i == codon_j) { sum_syn = sum_syn + (frequency_vector[i] * rate_matrix[i][j] * frequency_vector[j]); } }

} }

if (sum_all != 0) { rho_ns = (sum_all - sum_syn) / sum_all; rho_syn = sum_syn / sum_all; } else { rho_ns = 0; rho_syn = 0; }

}