GARN3 User Manual

GARN3 is a Java application and it requires Java version 17 or higher.

GARN3 uses Fasta file for primary/secondary structures. Example:

>1E8O
gGGCCGGGCGCGGUGGCGCGCGCCUGUAGUCCCAGCUACUCGGGAGGCUC
((((((((((((......)))))))....((((........)))))))))

Convert PDB file to GARN model

To create a file containing a GARN model from a PDB file: java -jar ./GARN3.jar PDBTOGARN ‹Name of molecule› ‹Fasta File› ‹Location of the PDB file›

Example:

java -jar ./GARN3.jar PDBTOGARN 1E8O 1E8O_fasta.txt 1E8O.pdb

Output two files : ‹Name of molecule›_GARN.csv with the position of nodes in 3D space, and ‹Name of molecule›_listNucleo.csv, to indicate the correspondance between nodes and nucleotides.

Predict molecules (run simulations)

To run the simualations and compute the sampling space: java -jar ./GARN3.jar GAME ‹Name of molecule› ‹Fasta File› ‹Number of solutions›

Example:

java -jar ./GARN3.jar GAME 1E8O 1E8O_fasta.txt 10

Output 10 files : GARN*‹Name of molecule›_0.csv, GARN*‹Name of molecule›1.csv, ..., GARN‹Name of molecule›_9.csv

If you wish to use an specific scoring algorithm for the molecule, please set the option EXP3 or UCB like this:

java -jar ./GARN3.jar GAME 1E8O 1E8O_fasta.txt 10 UCB

If you wish to compute the sampling space and also compute RMSD, you need to specify the GARN model file:

java -jar ./GARN3.jar GAME 1E8O 1E8O_fasta.txt 10 0 1E8O_GARN.csv

Please note that the "0" following the number of solutions means that we don't want to specify the scoring algorithm (it will be automatically chosen by the GARN algorithm). In this argument, ANY value different from "EXP3" and "UCB" will be ignored.

Output 11 files : GARN*‹Name of molecule›_0.csv, .. , & GARN*‹Name of molecule›_9.csv, ... and a ‹Name of molecule›_out.csv with the RMSD and the Maximum distance (to sort the sampling).

Calculate RMSD between predicted molecules

To compute the RMSD between GARN files (compare to one file): java -jar ./GARN3.jar GAME ‹Name of molecule› ‹GARN file of molecule 1› ‹GARN file of molecule 2› ...

Example:

java -jar ./GARN3.jar RMSD 1E8O 1E8O_GARN.csv GARN_1E8O_1.csv GARN_1E8O_2.csv GARN_1E8O_3.csv GARN_1E8O_4.csv

Output 1 file: ‹Name of molecule›_rmsd.csv with the RMSD between the first file and others files.

Calculate TM-Score between predicted molecules

To compute the TM-Score between GARN files (compare to one file): java -jar ./GARN3.jar TMSCORE ‹Name of molecule› ‹GARN file of molecule 1› ‹GARN file of molecule 2› ...

Example:

java -jar ./GARN3.jar TMSCORE 1E8O 1E8O_GARN.csv GARN_1E8O_1.csv GARN_1E8O_2.csv GARN_1E8O_3.csv GARN_1E8O_4.csv

Output 1 file: ‹Name of molecule›_tmscore.csv with the RMSD between the first file and others files.

Calculate minimum and maximum distance

To compute the maximum distance in structure and sort the sampling: java -jar ./GARN.jar SORT ‹Name of molecule› ‹GARN file of molecule 1› ‹GARN file of molecule 2› ...

Example:

java -jar ./GARN3.jar SORT 1E8O GARN_1E8O_1.csv GARN_1E8O_2.csv GARN_1E8O_3.csv GARN_1E8O_4.csv

Output 1 file: ‹‹Name of molecule››_sort.csv with the maximum distance for each solution.

Generate PDB/align/PyMOL outputs from GARN CSVs

To reproduce the GARNTOPDB flow:

java -jar ./GARN3.jar GARNTOPDB [‹Full PDB file› ‹listNucleo.csv file›] ‹game1.csv› [game2.csv ...]

Modes:

1. With PDB reference (using some full-atom experimental structure):

java -jar ./GARN3.jar GARNTOPDB 1E8O.pdb 1E8O_listNucleo.csv GARN_1E8O_0.csv GARN_1E8O_1.csv

2. Without PDB reference (first game file is used as reference):

java -jar ./GARN3.jar GARNTOPDB 1E8O_GARN.csv GARN_1E8O_0.csv GARN_1E8O_1.csv

Example:

java -jar ./GARN3.jar GARNTOPDB 1E8O.pdb 1E8O_listNucleo.csv 1E8O_GARN.csv GARN_1E8O_0.csv

Generated files:

1. GARN_*.pdb: PDB files generated from game CSVs.
2. <reference>.align: RMS alignment output for each generated structure against the selected reference.
3. If full reference mode is used: *_CG.pdb: reduced coarse-grained reference structure.
4. If full reference mode and PyMOL are available: *_CG.pse: optional PyMOL session file.

Contacts

When in doubt, you can contact jhonatansilva@usp.br or daniel.cordeiro@usp.br