DeepGCF learns the functional conservation between pig and human based on the epigenome profiles from two species. It can be applied to pairwise comparisons of any two species given their epigenome profiles sequence alignments.
DeepGCF scores for human (GRCh38) and pig (susScr11) are available here. In the human bed file, each window has a unique score, but in the pig bed file, there are overlapping windows (the same region may have multiple scores) because the input sequence alignment is generated using human genome as the basis, which could align to multiple segments of pig genome.
DeepGCF pipeline is mainly Python-based. Scripts involving data preparation and training DeepGCF are modified from LECIF. In order to download DeepGCF, you can clone this repository via the command
git clone https://github.com/liangend/DeepGCF.git
In order to install the Python dependencies, you will need the Anaconda package manager. After installing Anaconda, run the following commands to create an environment with DeepGCF's dependencies:
cd DeepGCF
conda env create --file deepgcf_env.yml
Once the above has completed, run the following command to activate the conda environment.
source activate deepgcf
The whole pipeline should be run under this environment to avoid errors caused by different package versions. See the following tutorials for detailed steps of DeepGCF.
3. Feature predition using DeepSEA
Li et al., Learning functional conservation between human and pig to decipher evolutionary mechanisms underlying gene expression and complex traits, Cell Genomics (2023), https://doi.org/10.1016/j.xgen.2023.100390