DeepGOMeta

This repository contains the scripts and datafiles used in the DeepGOmeta manuscript.

Dependencies

• The code was developed and tested using python 3.10.
• Clone the repository: git clone https://github.com/bio-ontology-research-group/deepgometa.git
• Create virtual environment with Conda or python3-venv module.
• Install PyTorch: pip install torch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2
• Install DGL: pip install dgl==1.1.2+cu117 -f https://data.dgl.ai/wheels/cu117/repo.html
• Install other requirements: pip install -r requirements.txt

Running DeepGOMeta model

Follow these instructions to obtain predictions for your proteins. You'll need around 30Gb storage and a GPU with >16Gb memory (or you can use CPU)

• Download the data.tar.gz
• Extract tar xvzf data.tar.gz
• Run the model python predict.py -if data/example.fa

Docker container

We also provide a docker container with all dependencies installed: docker pull coolmaksat/deepgometa
This repository is installed at /deepgometa directory. To run the scripts you'll need to mount the data directory. Example:
docker run --gpus all -v $(pwd)/data:/workspace/deepgometa/data coolmaksat/deepgometa python predict.py -if data/example.fa

Nextflow

DeepGOMeta can be run as a Nextflow workflow using the docker image for easier execution.

Requirements:

• For amplicon data: OTU table of relative abundance, where OTUs are classified using the RDP database
• For WGS data: Protein sequences in FASTA format

1. After cloning the repository, navigate to the Nextflow directory: cd Nextflow
2. Update the runOptions paths in nextflow.config
3. Navigate to the data directory cd data_and_scripts and download the genome annotations
4. Run workflow. Example: nextflow run DeepGOMeta.nf -profile docker/singularity --amplicon true --OTU_table otu_relative_abd.tsv --pkl_dir /PATH/TO/PKL/DIR/

Paired Datasets

1. Data and metadata: download from SRA and MG-RAST using sample accessions
2. Processing reads:
   • 16S reads - generate OTU tables using the Nextflow 16SProcessing workflow
   • WGS reads - obtain protein sequences using the assembly pipeline
3. Functional annotation:
   • OTU tables - generate a weighted functional profile for each OTU table using DeepGOmeta predictions
   • Protein fasta - run DeepGOmeta on Prodigal output from metagenome assemblies, and generate a binary functional profile for each dataset
4. Clustering and Purity: use a metadata file and the functional profile to apply PCA, k-means clustering, calculating purity, and generating plots for 16S datasets and WGS datasets
5. Information Content Calculation: create a .txt file for each sample containing the 16S predicted functions and WGS predicted functions on separate lines (e.g. 16Ssample'\t'GO1'\t'GO2'\n'WGSsample'\t'GO2'\t'GO3), and get IC for each function, then run a t-test