fsimkovic / CONFOLD2

Improved ab initio protein structure reconstruction

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool


CONFOLD version 2.0 (CONFOLD2) 7/12/2016

CONFOLD2 is a contact-guided protein structure prediction tool. Most scripts are written in Perl. It takes predicted contacts file (CASP RR format) and 3-state secondary structure (SCRATCH fasta format) as input and delivers top 5 models.


Installing CONFOLD

  1. Install CNS-suite (see below for instructions).
  2. Update paths for the variables $cns_suite in core.pl.
  3. Configure how CONFOLD2 will be parallelized. Current (default) configuration is to use HPC cluster. Make appropriate changes at lines 113-117 of confold2-main.pl to run in a local machine.
  4. Test the program:
$ ./confold2-main.pl -rr ./dry-run/input/1guu.rr -ss ./dry-run/input/1guu.ss -out ./output-1guu 

Installing CNS Suite

  1. To download CNS suite, provide your academic profile related information at http://cns-online.org/cns_request/. An email with (a) link to download, (b) login, and (c) password will be sent to you. Follow the link, possibly http://cns-online.org/download/, and download CNS suite cns_solve_1.3_all_intel-mac_linux.tar.gz.
  2. Unzip
$ tar xzvf cns_solve_1.3_all_intel-mac_linux.tar.gz
  1. Change directory to cns_solve
$ cd cns_solve_1.3
  1. Unhide the file .cns_solve_env_sh
$ mv .cns_solve_env_sh cns_solve_env.sh
  1. Edit (a) cns_solve_env.sh and (b) cns_solve_env to replace _CNSsolve_location_ with CNS installation directory. For instance, if your CNS installation path is /home/user/programs/cns_solve_1.3 replace _CNSsolve_location_ with this path
  2. Test CNS installation
$ source cns_solve_env.sh
$ cd test 
$ ../bin/run_tests -tidy *.inp

Predictions for PSICOV, CASP11, and CASP12 datasets

The contact predictions, secondary structure predictions, and CONFOLD2 reconstructed models for (a) PSICOV dataset using PSICOV contacts, (b) PSICOV dataset using MetaPSICOV contacts, (c) CASP11 dataset using CONSIP2 contacts, and (d) CASP12 dataset using RaptorX contacts are available at http://sysbio.rnet.missouri.edu/bdm_download/confold2/.


Contact

bap54@mail.missouri.edu (developer) chengji@missouri.edu (PI)

About

Improved ab initio protein structure reconstruction


Languages

Language:Perl 98.9%Language:Shell 1.0%Language:Scheme 0.1%