Experiments for MetaGeneMark-2

Georgia Institute of Technology, Atlanta, Georgia, USA

Reference: PAPER LINK

Overview

This repository contains the data and source code needed to reproduce all results found in the MetaGeneMark-2 paper.

Program Versions

MetaGeneMarKS is a standalone tool, but building the initial set of models relies on GeneMarkS-2 predictions. Similarly, results are compared to multiple external tools, whose versions are shown here:

• GeneMarkS-2:
• (Meta)Prodigal:
• MetaGeneAnnotator:
• FragGeneScan:
• MetaGeneMark:

MetaGeneMark-2 is a C++ program. That said, experiments and results are all executed and analyzed in python. To get all the packages (for reproducibility), it is recommended that the user creates a conda environment from the file in install/conda_mgm2.yml through the following command:

conda env create -f install/conda_mgm2.yml --name mgms

This can then be activate via

conda activate mgms

See info/reproduce.[html|pdf] for more information.

Installing MetaGeneMark-2 locally

Running MetaGeneMark-2 using automatic genetic code detection is done through the run_mgm.pl script found in $code/hmm_src. The below compiles the C++ binary and copies all the relevant components to $bin_external/mgm2_auto.

 cd code/hmm_src;
 pf_makefile=Makefile.macos    # NOTE: change based on operating system
 make -f $pf_makefile

This generates a binary =gmhmmp2=.

Running MetaGeneMark-2

Running MetaGeneMark-2 with automatic genetic code detection is done using run_mgm.pl. The following files should be in the same directory: run_mgm.pl, gmhmmp2, mgm2_11.mod, mgm2_4.mod. MetaGeneMark-2 can then be run (from anywhere) using:

$path_to_binary/run_mgm.pl --seq [name]  --out [name]

Required options:
     --seq  [name]            nucleotide sequence of metagenome in FASTA format.
     --out  [name]            output file with coordinates of predicted protein coding genes.

Output options:

       --nt  [name]           output file with nucleotide sequences of predicted genes in FASTA format.
       --aa  [name]           output file with protein sequences of predicted genes in FASTA format.
       --format  [gtf]        format of output file with gene coordinates: gtf or gff3.
       --clean                delete temporay files

Other parameters:
      --verbose

Reproducing Results

We provide a document detailing how to reproduce all results. This can be found at info/reproduce.[html|pdf]

Folder structure

The following directory structure should be maintained while running experiments

.
├── bin                                   # Executables constructed from python/bash drivers (via install.sh)
├── bin_external                          # External tools
├── config                                # Configuration files, e.g. MetaGeneMark-2 learning parameters
├── config.sh                             # Load bash variables for paths to different directories
├── install                               # Conda environment file for easy installation
├── lists                                 # Lists of genomes (main input method to scripts)
├── info                                  # Information about reproducing results
├── metadata                              # Non-genomic data, including taxonomy information
├── data                                  # Data Location: where all raw data will be stored during runs
│   ├── GCFID 1                           # ID of genome 1
│   │   ├── ncbi.gff                      # RefSeq annotation
│   │   ├── sequence.fasta                # Genomic sequence file
│   ├── GCFID 2                           # ID of genome 2
│   │   ├── ncbi.gff                      # RefSeq annotation
│   │   ├── sequence.fasta                # Genomic sequence file
│   │   ...
├── code                                  # Source code
│   ├── python                            # Python code
│   │   ├── driver                        # Drivers that can be executed
│   │   ├── lib                           # Library files
|   |── mgms                              # MetaGeneMarkS (C++) source code and Makefile
│   ├── bash                              # Bash scripts
│   │   ├── driver                        # Drivers that can be executed
│   │   ├── lib                           # Library files
├── runs                                  # Data Location: where all raw data will be stored during runs
│   ├── GCFID 1                           # ID of genome 1
│   │   ├── startlink                     # StartLink runs
│   │   ├── mgms                          # MGMS runs
|   |   ├── others...                     # Other tools
│   ├── GCFID 2                           # ID of genome 2
│   │   ├── startlink                     # StartLink runs
│   │   ├── mgms                          # MGMS runs
|   |   ├── others...                     # Other tools
│   │   ...