EcoFoldDB

KO | EN

gitlite — search

#javascript #hacktoberfest #react #llm #android #machine-learning #linux #macos #java #awesome #cli #vue #mcp #open-source

(★ 34)

Database and pipeline for protein structure-guided annotations of ecologically relevant functions at the metagenome scale.

EcoFoldDB 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# EcoFoldDB
Database and pipeline for protein structure-guided functional profiling of ecologically relevant microbial traits at the metagenome scale (millions of proteins).

EcoFoldDB is a database of protein structures encoded by microbial genes involved in:  
Trace gas oxidation, carbon cycling (i.e., C fixation, metabolism and degradation), nitrogen cycling, sulphur cycling, phosphorus cycling, iron cycling, plant-microbe interactions, and osmotic stress tolerance.

`EcoFoldDB-annotate` is an annotation pipeline using [Foldseek](https://doi.org/10.1038/s41587-023-01773-0) for structure-based functional annotations against EcoFoldDB at the metagenome-scale. Input can be either raw protein sequences (in which case the pipeline leverages the [ProstT5](https://doi.org/10.1093/nargab/lqae150) protein language model so that you don’t need to run structure prediction) or an already built Foldseek database of protein structures.

# Publication
[Ghaly, T.M., Rajabal, V., Russell, D., Colombi, E. and Tetu, S.G (2025) EcoFoldDB: Protein structure-guided functional profiling of ecologically relevant microbial traits at the metagenome scale. *Environmental Microbiology*, 27: e70178; doi: https://doi.org/10.1111/1462-2920.70178](https://doi.org/10.1111/1462-2920.70178)

# Installation

### Clone EcoFoldDB repository
```
git clone https://github.com/timghaly/EcoFoldDB.git
cd EcoFoldDB
chmod +x EcoFoldDB-annotate.sh
```
### Install Foldseek
`EcoFoldDB-annotate` requires Foldseek for protein annotations.

If using protein sequences as input, then a GPU-compatible Foldseek is highly recommended for metagenome-scale annotations (i.e., for millions of proteins). Foldseek converts protein sequences directly to structural representations using the ProstT5 protein language model. GPU can accelarate ProstT5 inference by one to two orders of magnitude.

If using protein structures as input (must be converted to a Foldseek database), then either a CPU-only or GPU-Foldseek will work equally.

There are different options to install Foldseek:

*Option 1 (GPU Foldseek): Download Foldseek's precompiled binary - Linux AVX2 & GPU build*
```
wget https://mmseqs.com/foldseek/foldseek-linux-gpu.tar.gz
tar xvfz foldseek-linux-gpu.tar.gz
```
*Option 2 (GPU Foldseek): Instead, compile GPU Foldseek binary from source to be optimised to your specific system*:

```
conda create -n nvcc -c conda-forge cuda-nvcc cuda-cudart-dev libcublas-dev libcublas-static cuda-version=12.6 cmake
conda activate nvcc
git clone https://github.com/steineggerlab/foldseek.git
cd foldseek
mkdir build && cd build
cmake -DCMAKE_BUILD_TYPE=RELEASE -DCMAKE_INSTALL_PREFIX=. -DENABLE_CUDA=1 -DCMAKE_CUDA_ARCHITECTURES="native" ..
make -j8
make install
```

*Option 3 (CPU-only Foldseek): Other Foldseek installation options*:

For precompiled CPU-only Foldseek binary, or conda installation instructions, see the [Foldseek installation page](https://github.com/steineggerlab/foldseek?tab=readme-ov-file#installation).

### Download ProstT5 protein language model (If input are protein sequences)

If using protein sequences as input, `EcoFoldDB-annotate` requires the ProstT5 model to be locally installed.  
The model can be downloaded using Foldseek:

```
foldseek databases ProstT5 ProstT5_dir tmp --remove-tmp-files 1
```

# Usage
If Foldseek is not in your `$PATH`, then you can add its binary directory to the `PATH` environment variable before running `EcoFoldDB-annotate.sh`:
```
export PATH=/full/path/to/foldseek/bin/:$PATH
```
Or instead, the foldseek binary directory can be provided to `EcoFoldDB-annotate.sh` using the flag:  
``` --foldseek_bin ```

```
./EcoFoldDB-annotate.sh --EcoFoldDB_dir PATH [--gpu (0|1)] [--ProstT5_dir PATH] [--foldseek_bin PATH] [-e EVALUE] [--qcov QCOV] [--tcov TCOV] [--tmp-dir PATH] [--remove-tmp-files (0|1)] [-o OUTDIR] INPUT_FILE_OR_DB
EcoFoldDB-annotate v2.1.0

Input types:
  FASTA file:        File with .fasta, .fa, or .faa extension
  Foldseek database: Path to and including database name

Mandatory parameters:
  --EcoFoldDB_dir    Full path to EcoFoldDB_v2.0 directory
  INPUT_FILE_OR_DB   Input FASTA file of protein sequences OR Foldseek structural database

Mandatory parameters required for FASTA input:
  --ProstT5_dir      Full path to ProstT5 model directory (required for FASTA input)
  --gpu              Use GPU (1) or CPU (0) (required for FASTA input)

Optional parameters:
  --foldseek_bin     Path to directory containing foldseek binary
  -e                 E-value threshold (default: 1e-10)
  --qcov             Minimum query coverage (default: 0.8)
  --tcov             Minimum target coverage (default: 0.8)
  --tmp-dir          Temporary directory for Foldseek (default: OUTDIR/results_db/NAME_tmp)
  --remove-tmp-files Remove temporary files (0=no, 1=yes) (default: 0)
  -o                 Output directory to be created (default: EcoFoldDB_annotate)
  -h, --help         Show this help message
  --version          Show version information

```
# Output
The main annotation results will be in the output directory with the file extension `.ecofolddb_annotations.txt`

### Output directory structure
```
OUTDIR/
├── ${name}.ecofolddb_annotations.txt                       # Main annotation result
├── Filtered_seqs/
│   ├── ${name}.length_filtered.fasta                       # Input protein sequences included after filtering
│   └── ${name}.excluded_long_seqs.fasta                    # Input protein sequences excluded after filtering
├── ProstT5_db/
│   └── ${name}_db*                                         # ProstT5 Foldseek database files (binary database files)
└── results_db/                                             
    ├── ${name}_results.*                                   # Foldseek result DB files produced by foldseek `search` command
    ├── ${name}_foldseek_results.txt                        # All Foldseek hits to target EFDB and non-target Swiss-Prot structures
    ├── ${name}_foldseek_results.top_hits.txt               # Top hit for each query protein
    ├── ${name}_foldseek_results.top_EFDB_hits.txt          # Queries whose top hit is with a target EFDB structure
    ├── ${name}_foldseek_results.top_EFDB_hits.Filtered.txt # Top EFDB hits filtered by alignment coverage e-value thresholds
    └── valid_targets.txt                                   # Target EFDB structures from EcoFoldDB
```

# Citations
If you have used EcoFoldDB, please cite the following:  
[Ghaly, T.M., Rajabal, V., Russell, D., Colombi, E. and Tetu, S.G (2025) EcoFoldDB: Protein structure-guided functional profiling of ecologically relevant microbial traits at the metagenome scale. *Environmental Microbiology*, 27: e70178; doi: https://doi.org/10.1111/1462-2920.70178](https://doi.org/10.1111/1462-2920.70178)

Please also cite the following dependencies:

Foldseek:  
[van Kempen M, Kim S, Tumescheit C, Mirdita M, Lee J, Gilchrist CLM, Söding J, and Steinegger M. (2023) Fast and accurate protein structure search with Foldseek. *Nature Biotechnology*, doi:10.1038/s41587-023-01773-0](https://doi.org/10.1038/s41587-023-01773-0)  
If you have used ProstT5:  
[Heinzinger, M., Weissenow, K., Sanchez, J.G., Henkel, A., Mirdita, M., Steinegger, M., and Rost, B. (2024) Bilingual language model for protein sequence and structure, *NAR Genomics and Bioinformatics*, doi:10.1093/nargab/lqae150](https://doi.org/10.1093/nargab/lqae150)

EcoFoldDB

EcoFoldDB 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

EcoFoldDB

EcoFoldDB 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README