/evoAllostery

Statistical coupling analysis of siglec proteins

Primary LanguagePython

Graphical Summary of README

Statistical Coupling Analysis

Table of Contents

  1. Overview
    • Introduction
    • Raw data description
  2. Installation
    • Folder structure
    • Installing dependencies
  3. Multiple sequence alignment
    • Alignment protocols
    • Processing
  4. Generate Analysis
    • data_analysis.py

I. Overview

Introduction

The purpose of this repo is to facilitate running statistical coupling analysis for the V-set family proteins. Specifically, the project is looking at amino acid co-evolution in human siglecs. The repo starts from raw structural and sequence data and generates a multiple sequence alignment based on the sequence of Siglec-7. The alignment is then processed to extract evolutional information to find domains under unique selectional pressure. The pySCA code is written and maintained by the Ranganathan Lab and is used extensively throughout the analysis (1,2). It is documented here:
https://ranganathanlab.gitlab.io/pySCA/

Raw data description

Custom MSA

The multiple sequence alignment was generated using PSI-BLAST, with the sequence of the siglec-7 crystal structure used as the query (PDB:1O7V). To increase the accuracy and prevent false hits, we set the e-value threshold to 1E-7 and a word size of two in both iterations. A total of 9552 sequences were retrieved from the non-redundant (nr) database. For phylogenetic annotation, we wrote a script to parse the accession numbers from the text file (can be found in the Scripts directory). We used the 'ncbi' option in annotateMSA, which uses the Biopython NCBI Entrez utility to look up taxonomic information.

PFAM Analysis

This file contains a multiple sequence alignment generated from 73,704 V-set sequences. Using the pySCA API, the multiple sequence alignment has been enhanced to include phylogenetic metadata. This dataset can be easily accessed via the PFam database website. The annotation step is computationally expensive can take up to 7 hours to download and annotate. For the original multiple sequence alignment data visit the following site and select Generate:
https://pfam.xfam.org/family/V-set#tabview=tab3

II. Installation

Folder structure

The repo is divided into three key directories: Data, Figures, and Scripts. The Data directory contains three sub folders Alignments, Fasta, and Structures. Alignments contains all six of the multiple sequence alignments we used in the paper. Fasta contains the original psi-blast file and Structures contains the final PyMol files to vizualize the sectors. The Figures directory contains is where any output figures will be saved. The Scripts directory contains our full python script called data_analysis that interacts with the processed alignment file. It also contains the R script that was used to perform the Decipher multiple sequence alignment. The script called data_analysis currently will generated a single figure.

Installing dependencies

In order to run these calculations, we will need to install some dependencies. First, we will need to install the core dependencies which include the latest versions of Python and GCC. The Ranganathan Lab has included a beautiful tutorial with the pySCA distribution. Completing the Ranganathan Lab pySCA installation tutorial will automatically install several other dependencies, including: numpy, scipy, argparse, wheel, and matplotlib. You can use a package manager like homebrew to keep the installation clean. The installation is summarized below as a supplement to the Ranganathan tutorial. However, make sure to refer to the original tutorial for system specific details details: https://ranganathanlab.gitlab.io/pySCA/install/

  1. Core system dependencies

    xcode-select --install
    brew install python3
    brew install gcc

  2. Fasta36

    git clone https://github.com/wrpearson/fasta36.git
    cd fasta36/src
    make -j2 -f ../make/Makefile.os_x86_64 all
    cp -r ../bin /usr/local
    rm /usr/local/bin/README
    cd ../..

  3. pySCA

    git clone https://github.com/ranganathanlab/pySCA.git
    cd pySCA-master
    pip install .

  4. The pySCA distribution automatically installs the latest compatible versions of the following packages. If you have questions about versions, visit the pySCA GitHub for more information as they change from time to time.

    • numpy
    • scipy
    • argparse
    • wheel
    • matplotlib

III. Custom Multiple Sequence Alignment

Alignment protocols

To facilitate reproducibility, we have included all six multiple sequence alignment test cases referenced in the paper. The alignment files were generated with Decipher, Mafft L-INS-i, MUSCLE, Kalign, FAMSA, or Promals3D. Depending on the algorithm, the processing time can range anywhere from a few seconds to a few days. Therefore, to facilitate reproducibility, we have included all processed alignments. However, once your alignment program of choice has been downloaded, you can generate alignments with the following commands:

Decipher: R script
L-INS-i: mafft-linsi --localpair --anysymbol --thread -1 --treeout in.fasta > out.an
MUSCLE: ./muscle3.8.31_i86darwin64 -in v9.fasta -out v9.an
Kalign: kalign -i in.fasta -o out.an
FAMSA: ./famsa in.fasta out.an
Promals3D: http://prodata.swmed.edu/promals3d/promals3d.php

Processing

Add the alignment file and the PDB to the pySCA directory at ./pySCA-master/data/. Run the following commands from the terminal to process the annotated MSA file. These commands create a database file (.db) that will be used by pySCA. For benchmarking purposes, you should know that the first line is the most computationally expensive. It can take several hours. The following three should take less than 10 minutes.

annotateMSA -i ../data/PF07686_full.txt -o ../data/PF07686_full.an -a 'pfam' -p ./data/pfamseq.txt
scaProcessMSA -a ../data/PF07686_full.an -b ../data/ -s 3BIK -c B -f 'Homo sapiens' -t -n
scaCore -i ../output/PF07686_full.db
scaSectorID -i ../output/PF07686_full.db

IV. Generating the correlation matrix

Download the siglec_sca repository. Inside the scripts directory of the siglec_sca Github repository, you will find a python script. It takes a single file as input, the .db file generated as described above. With pySCA and dependencies successfully installed, this script will create a all the figures referenced in our paper that can be generated programatically. It will also generate your PyMol files for vizualization.

git clone https://github.com/davidkastner/siglec-sca.git
cd domain-clusters/Scripts
python data_analysis.py

Citations

  1. Rivoire, O., et al. Evolution-Based Functional Decomposition of Proteins. (2016)
  2. Halabi, N., et al. Protein sectors: evolutionary units of three-dimensional structure. Cell. 138, (2009)
  3. Salinas, V., et al. Coevolution-based inference of amino acid interactions underlying protein function (2018)