Microorganism identification could be misleading or slow with traditional methods (16S, phenotypical analysis).
This repository contains the scratch of a pipeline for alternative classification using codon usage patterns.
Here you will find a pipeline to generate codon usage signature tables to train an ML model, starting from codon abundance in CDS.
Clone the repo containig the scripts and run it as follows:
git clone https://github.com/apollo994/codon_signature.git
bash codon_signature/preprocessing/codon_signature.sh RefSeq_table Taxon_IDs_list run_name
Where RefSeq_table is a table containing the following columns (tab-separated):
- Taxid : the specific ID for the species in RefSeq
- Species : scientific name of the species
- Assembly : unique identifier of the deposited genome assembly
- AAA...TTT : 64 columns representing codons and their absolute abundance in the CDS of the assembly
Taxon_IDs_list is a .txt file that contains one Taxid per line and represents the list of species you want to generate the final codon signature table for (one for each Taxid)
run_name is the name used to name the folder containing as many files as Taxids you asked for, each of them representing the codon signature of each species.