Theory of measurement for site-specific evolutionary rates in amino-acid sequences

This repository contains all the scripts and data to reproduce the results of:

D. K. Sydykova, C. O. Wilke (2018). Theory of measurement for site-specific evolutionary rates in amino-acid sequences

Contents

• analytical_rates contains rates that were calculated using analytical derivations. The following list is the files contained in this direcotry and their descriptions.

  • all_sites_aa.csv contians site-wise rates for every site in egg white lysozyme (PDB ID: 132L) calculated for different times (column site in all_sites_aa.csv directly corresponds to column SITE in 132L_A_foldx_ddG.txt). These rates were calculated assuming that the true model is a mutation-selection model, and the inference model is Jukes-Cantor (equations 3-5). This file was generated with the command

    python analytical_rate_aa.py -m 125 -q q_matrix/amino_acid/ -o all_sites_aa.csv.

  • ten_sites_aa.csv contains site-wise rates for the first ten sites in egg white lysozyme (PDB ID: 132L) calculated for different times (column site in ten_sites_aa.csv directly corresponds to column SITE in 132L_A_foldx_ddG.txt). These rates were calculated assuming that the true model is a mutation-selection model, and the inference model is Jukes-Cantor (equations 3-5). This file was generated with the command

    python analytical_rate_aa.py -m 10 -q q_matrix/amino_acid/ -o ten_sites_aa.csv.

  • ten_sites_aa_true_JC.csv contains site-wise rates for ten sites that were calculated under the assumption that both the true model and the inference model are Jukes-Cantor. This file was generated with the command

    python analytical_rate_aa_true_JC.py -m 10 -o ten_sites_aa_true_JC.csv.

  • ten_sites_aa_QM.csv contains site-wise rates for when rate is measured with an arbitrary Q_M matrix and for when rate is measured with a Jukes-Cantor matrix (equation 1).

  • ten_sites_codon.csv contians site-wise rates for every site in egg white lysozyme (PDB ID: 132L) calculated for different times (column site in all_sites_aa.csv directly corresponds to column SITE in 132L_A_foldx_ddG.txt). These rates were calculated assuming that the true model is a codon mutation-selection model, and the inference model is an amino acid Jukes-Cantor (equation 6 and equations 22S and 24S). This file was generated with the command

    python analytical_rate_codon.py -m 10 -q q_matrix/codon/ -o ten_sites_codon.csv.

• inferred_rates contains files with site-wise rates inferred with HyPhy. There are two directories in inferred rates, raw_rates and processed_rates. raw_rates contains individual files for a simulated alignment (one file per alignment), and processed_rates contains concatenated files from raw_rates.The following list describes the directories contained in raw_rates:

  • JC, inferred rates when the true and the inference models are both Jukes-Cantor-like.
  • all_sites, inferred rates when the true model is MutSel, and the inference model is either Jukes-Cantor-like (JC), WAG, JTT, or LG.
  • site_dupl, rates inferred with JC for alignments with different number of site duplicates.
  • ten_sites, inferred rates when the true model is MutSel, and the inference model is JC.
  • translated, inferred rates when the true model is a codon MutSel model, and the inference model is amino acid JC.

• q_matrix contains site-wise substitution matrices Q used for simulating alignments and for calculating site-wise rates. There are two directories in q_matrix, amino_acid and codon. amino acid contains amino acid substitution matrices, and codon contains codon substitution matrices.

  Files that start with 132L_A indicate substitution matrices that were calculated using data from Echave et al (2015) for egg white lysozyme (PDB ID: 132L). For example, file 132L_A_site79_q_matrix.npy corresponds to the substitution matrix calculated for site 79. The site positions here correspond to the site positions given by the file 132L_A_foldx_ddG.txt, which was directly copied from the git repository for Echave et al (2015) https://github.com/wilkelab/therm_constraints_rate_variation. These matrices were calculated according to the mutation-selection (MutSel) theory by Halpern and Bruno (1998). The script to calculate amino acid MutSel matrices is src/calculate_aa_mutsel_Q.py, and the script to calculate codon MutSel matrices is src/calculate_codon_mutsel_Q.py.

  Files that start with site0_JC indicate substitution matrices Q defined as Q = r^(k)Q_JC. Here, Q_JC is the Jukes-Cantor-like matrix, and r^(k) is the true rate at site k. True rates were generated by the script src/analytical_rate_aa_true_JC.py and were stored in analytical_rates/ten_sites_aa_true_JC.csv. The script to calculate amino acid substitution matrices Q defined as Q = r^(k)Q_JC is src/calculate_aa_true_JC_Q.py.

• trees contains tree files that were used for simulating alignments. Each file name stores the number of branches (n) in the tree and the branch lengths (bl). For example, file n2_bl0.005.tre describes a tree with 2 branches of lengths 0.005 each.

• hyphy contains all scripts and files to run HyPhy.

• plots contains plots generated for the manuscript.

• src contains code to run the analysis.

Analysis