arupmondal835 / NFold

Process Machine Learning Histogram data to setup MELD folding simulations

NFold

Process Machine Learning Histogram data to setup MELD folding simulations

Instruction for running Meld simulation

04/30/2021 PerezLab@UF

---

Required software:

Amber: http://ambermd.org/ OpenMM: https://github.com/openmm/openmm Meld: https://github.com/maccallumlab/meld

---

step 1: Generate informational distograms, phi, psi from corresponding .npy prediction file.

python analyze_distograms.py

Description: Select high informational range of distograms, psi, phi from enormous amount of probability histogram predictions. Input: distogram.npy, phi.npy, psi.npy, sequence.fa
Output: contacts.dat, tight_contacts.dat, phi.dat, tight_phi.dat, psi.dat, tight_psi.dat

step 2: Generate starting structure from Amber minimization.

sbatch setup_from_random_cpu.sh

Description: Before starting Meld simulation, we generate a starting system from given sequence and then minimize it with Amber. Input: sequence.dat (contains only the second line of sequence.fa file) Output: TEMPLATES/minimized.pdb

step 3: Start Meld simulation with OpenMM.

sbatch job_30_new.sh (contains 'python setup_aMeld.py' inside)

Description: Setup Meld simulation using ff14SBonlysc force field + restraints derived from output files in step 1. Input: setup_aMeld.py, sequence.dat, phi.dat, tight_phi.dat, psi.dat, tight_psi.dat, contact.dat, tight_contact.dat Output: Data/ remd.log (Data/ stores all simulation output, remd.log keep track of the simulation.)