This is the repo for the Orthology Inference project. Its aim is to identify orthologous groups of proteins across 33 Drosophila genomes. Like many other orthology inference pipelines, we use a graph-based method of clustering reciprocal best hits from BLAST searches between pairs of genomes. The chief innovations, however, are 1) the extension of commonly used triangle clustering to its generalization of k-clique percolation and 2) the implementation of two phylo-HMMs to identity poorly supported regions and missing data in the alignments. These methods and their results are discussed in further detail in the accompanying pre-print article.
At the highest level, this project is organized into the following components:
orthology_inference/
├── analysis/
├── bin/
├── data/
├── src/
└── README.md
Only analysis/ and src/, which together contain all code written for this project, are explicitly tracked by Git. bin/ contains third-party programs or code used in this project. Though this directory is not tracked by Git, scripts may reference it by path, so it is included here for completeness. Similarly, data/, which contains all the raw data used in this project, is not tracked by Git.
analysis/ contains only directories, which serve to group related analyses. Some directories are "orphaned" and no longer contribute to any recent or ongoing analyses, but are included here for completeness. Currently, it contains the following entries:
ortho_cluster1/: Construction of orthologous groups using all Drosophila genome annotations obtained from NCBIortho_cluster2/: Final set of orthologous groups removing genomes which clustered poorly inortho_cluster1/ortho_MSA/: Creation of multiple sequence alignments from orthologous groups generated inortho_cluster2/ortho_tree/: Set of orthologous groups using all genomes inortho_cluster2/and an outgroup S. lebanonensis; this directory exists solely to calculate a phylogenetic tree for the species inortho_cluster2/ortho_search/: Scripts to run and parse BLAST searches for all genomes used inortho_cluster1/
Nearly all code is written in Python and run with version 3.9.12. The remaining code is written in Bash shell scripts. The following Python libraries were used.
| Library | Version |
|---|---|
| homomorph | 0.3.0 |
| matplotlib | 3.5.1 |
| NetworkX | 2.7.1 |
| NumPy | 1.22.3 |
| pandas | 1.4.1 |
| SciPy | 1.9.1 |
| scikit-bio | 0.5.7 |
Scikit-Bio attempts to import some deprecated warnings from scipy.stats during import, so these lines were commented out to ensure compatibility.