https://github.com/cokelaer/defi-bioinfo

Statistical exploration of codon usage

https://github.com/cokelaer/defi-bioinfo

Repository

Statistical exploration of codon usage

https://github.com/cokelaer/Defi-Bioinfo/blob/main/

# Statistical exploration of codon usage

This project is part of "Dfi-Bioinformatique" at INSA/ENSAT Toulouse, France, consisting of statistical exploration of start and stop codons in prokaryotes. The objective is to know the distribution of these codons according to the bacterial phyla and to allow the development of a prediction tool that, from an unknown genome, would assign the corresponding phylum based on the proportions of its start and stop codons.

Firstly, the data collected from [NCBI](https://www.ncbi.nlm.nih.gov/nuccore) are preprocessed using the following command, that allows us to read the fasta/gff files, extract codons and convert raw data into DataFrame (assume that this process has already been done in this git and the generated DataFrames are available at `./data`).
```bash
python data_preprocessing.py
```
```
 Preprocessing gff at ../raw_data\actinobacteria.gff3
 Preprocessing fasta at ../raw_data\actinobacteria.fasta
    Extracting codons: 100%|| 75/75 [03:13<00:00,  2.58s/it]
 Preprocessing gff at ../raw_data\cfb.gff3
 Preprocessing fasta at ../raw_data\cfb.fasta
    Extracting codons: 100%|| 75/75 [02:45<00:00,  2.21s/it]
 Preprocessing gff at ../raw_data\cyanobacteria.gff3
 Preprocessing fasta at ../raw_data\cyanobacteria.fasta
    Extracting codons: 100%|| 75/75 [03:06<00:00,  2.49s/it]
 Preprocessing gff at ../raw_data\firmicutes.gff3
 Preprocessing fasta at ../raw_data\firmicutes.fasta
    Extracting codons: 100%|| 75/75 [02:07<00:00,  1.70s/it]
 Preprocessing gff at ../raw_data\fusobacteria.gff3
 Preprocessing fasta at ../raw_data\fusobacteria.fasta
    Extracting codons: 100%|| 75/75 [01:32<00:00,  1.23s/it]
 Preprocessing gff at ../raw_data\proteobacteria.gff3
 Preprocessing fasta at ../raw_data\proteobacteria.fasta
    Extracting codons: 100%|| 75/75 [02:15<00:00,  1.81s/it]
 Preprocessing gff at ../raw_data\spirochetes.gff3
 Preprocessing fasta at ../raw_data\spirochetes.fasta
    Extracting codons: 100%|| 75/75 [01:11<00:00,  1.05it/s]
```

The main script is written in `main.py` involving statistical visualization, cluster analysis and classification. Run this code in a terminal with the following command:
```bash
python main.py
```
```
 Linear Classification
 Scores of prediction based on start codon:
    Phylum Actinobacteria: 0.9375
    Phylum CFB: 1.0
    Phylum Proteobacteria: 0.0
    Phylum Firmicutes: 0.7692307692307693
    Global score: 0.6333333333333333
 DecisionTree Classification
 Scores of prediction based on start codon:
    Phylum Actinobacteria: 0.9375
    Phylum CFB: 0.7692307692307693
    Phylum Proteobacteria: 0.5555555555555556
    Phylum Firmicutes: 0.9230769230769231
    Global score: 0.7833333333333333
 KNeighbors Classification
 Scores of prediction based on start codon:
    Phylum Actinobacteria: 0.9375
    Phylum CFB: 1.0
    Phylum Proteobacteria: 0.9444444444444444
    Phylum Firmicutes: 0.7692307692307693
    Global score: 0.9166666666666666
 RandomForest Classification
 Scores of prediction based on start codon:
    Phylum Actinobacteria: 0.9375
    Phylum CFB: 0.9230769230769231
    Phylum Proteobacteria: 0.9444444444444444
    Phylum Firmicutes: 0.9230769230769231
    Global score: 0.9333333333333333
 SVM Classification
 Scores of prediction based on start codon:
    Phylum Actinobacteria: 0.875
    Phylum CFB: 1.0
    Phylum Proteobacteria: 0.9444444444444444
    Phylum Firmicutes: 0.6153846153846154
    Global score: 0.8666666666666667
 NeuralNetwork Classification
 Scores of prediction based on start codon:
    Phylum Actinobacteria: 0.9375
    Phylum CFB: 0.9230769230769231
    Phylum Proteobacteria: 0.9444444444444444
    Phylum Firmicutes: 0.7692307692307693
    Global score: 0.9
```
The output figures can be found in `./figs`.

Owner

GitHub Events