Project Name: DNA Sequence Classification

Overview

Recent studies in DNA sequence classification have leveraged sophisticated machine learning techniques, achieving notable accuracy in categorizing complex genomic data. Methods such as k-mer counting have proven effective in distinguishing sequences from varied species like chimpanzees, dogs, and humans. These methods have been used widely in the latest literaure. However, these approaches often demand extensive computational resources. Our study introduces a novel adaptation of Jiang et al.'s compressor-based, parameter-free classification method, specifically tailored for DNA sequence analysis. This approach not only aligns with the current state-of-the-art in terms of accuracy but also offers a more resource-efficient alternative to traditional machine learning methods. A list of different algorithms was used respectively and our results demonstrate the proposed methods' comparable effectiveness in classifying DNA sequences.

Quick Start

git clone https://github.com/sukruozan/dna-sequence-classification.git

Environment

Conda is a cross platform package and environment manager that installs and manages packages. You can use it to replicate the same environment by simply running the following:

conda env create -f environment.yml

Dataset

• The original dataset can be reached at this Kaggle post Demystify DNA Sequencing with Machine Learning
• Here I combined all the data and created a single training and test datasets. For reproduction of the same results you can find the corresponding dataset in the folder dataset of this repository.

Dataset Composition by Species

Species	Train Size	Test Size
Chimpanzee	1345	337
Human	3504	876
Dog	656	164
Total	5505	1377

Distribution of Gene Family Samples by Species

Gene Family	Class Label	Chimpanzee	Human	Dog	Total
G protein coupled receptors	0	234	531	131	896
Tyrosine kinase	1	185	534	75	794
Tyrosine phosphatase	2	144	349	64	557
Synthetase	3	228	672	95	995
Synthase	4	261	711	135	1107
Ion channel	5	109	240	60	409
Transcription factor	6	521	1343	260	2124

Class Distribution - All Species	Class Distribution - Human	Class Distribution - Chimpanzee	Class Distribution - Dog

Confusion matrices for subspecies classifications: Human, Chimpanzee, and Dog DNA. These matrices detail the classifier's accuracy for each subspecies, highlighting the precision in distinguishing between these specific genomic sequences. The outcomes from the experiments performed by using brotli compressor is selected for depiction.

Comparison of Compression Algorithms in DNA Sequence Classification

Algorithm	Computation Time (seconds)	Accuracy	Recall	Precision	F1 Score
Gzip	1735.81	0.962	0.962	0.963	0.962
Snappy	1726.49	0.932	0.932	0.933	0.932
Brotli	12551.60	0.966	0.966	0.967	0.966
LZ4	1618.27	0.942	0.942	0.943	0.942
Zstandard	1560.72	0.930	0.930	0.935	0.931
BZ2	2657.35	0.924	0.924	0.924	0.924
LZMA	9486.67	0.958	0.958	0.958	0.958

Performance Metrics of Compression Algorithms on DNA Sequence Classification for Different Species

Algorithm	Chimpanzee Accuracy	Chimpanzee Precision	Chimpanzee Recall	Chimpanzee F1 Score	Human Accuracy	Human Precision	Human Recall	Human F1 Score	Dog Accuracy	Dog Precision	Dog Recall	Dog F1 Score
Gzip	0.997	0.997	0.997	0.997	0.944	0.945	0.944	0.944	0.988	0.989	0.988	0.988
Snappy	0.994	0.994	0.994	0.994	0.943	0.944	0.943	0.943	0.744	0.748	0.744	0.745
Brotli	1.000	1.000	1.000	1.000	0.947	0.950	0.947	0.948	0.994	0.994	0.994	0.994
LZ4	0.997	0.997	0.997	0.997	0.944	0.946	0.944	0.944	0.817	0.829	0.817	0.819
Zstandard	0.994	0.994	0.994	0.994	0.900	0.909	0.900	0.901	0.963	0.967	0.963	0.963
BZ2	0.985	0.985	0.985	0.985	0.893	0.894	0.893	0.893	0.963	0.966	0.963	0.964
LZMA	1.000	1.000	1.000	1.000	0.936	0.937	0.936	0.936	0.988	0.989	0.988	0.988

Confusion Matrix - All Species	Confusion Matrix - Human	Confusion Matrix - Chimpanzee	Confusion Matrix - Dog

Confusion matrices for subspecies classifications: Human, Chimpanzee, and Dog DNA. These matrices detail the classifier's accuracy for each subspecies, highlighting the precision in distinguishing between these specific genomic sequences. The outcomes from the experiments performed by using brotli compressor is selected for depiction.

K-Mer distribution in Class 6 sequences of Human genome

Support

If you need help or have a question, raise an issue or contact me at sukruozan@gmail.com.

Contributors

• Sukru Ozan
• Download the manuscript from arXiv To cite this work in your publications, use the following BibTeX entry:

Citation

  @article{ozan2024dna,
        title={DNA Sequence Classification with Compressors}, 
        journal={arXiv},
        author={Şükrü Ozan},
        year={2024},
        eprint={2401.14025},
        archivePrefix={arXiv},
        primaryClass={q-bio.GN},
        doi={https://doi.org/10.48550/arXiv.2401.14025},
        }

References

1. Jaddi, N. S., & Saniee Abadeh, M. (2022). "Cell separation algorithm with enhanced search behaviour in miRNA feature selection for cancer diagnosis." Information Systems, 104, 101906. DOI
2. Khan, S., Khan, M., Iqbal, N., Li, M., & Khan, D. M. (2020). "Spark-Based Parallel Deep Neural Network Model for Classification of Large Scale RNAs into piRNAs and Non-piRNAs." IEEE Access, 8, 136978–136991. DOI
3. Yagin, F. H., et al. (2023). "Explainable artificial intelligence model for identifying COVID-19 gene biomarkers." Computers in Biology and Medicine, 154, 106619. DOI
4. Wen, J., et al. (2019). "A classification model for lncRNA and mRNA based on k-mers and a convolutional neural network." BMC Bioinformatics, 20(1). DOI
5. Millán Arias, P., et al. (2022). "DeLUCS: Deep learning for unsupervised clustering of DNA sequences." PLOS ONE, 17(1), e0261531. DOI
6. Bentley, J. L., et al. (1986). "A locally adaptive data compression scheme." Communications of the ACM, 29(4), 320–330. DOI
7. Burrows, M. (1994). "A block-sorting lossless data compression algorithm." SRS Research Report, 124.
8. Huffman, D. (1952). "A Method for the Construction of Minimum-Redundancy Codes." Proceedings of the IRE, 40(9), 1098–1101. DOI
9. Ziv, J., & Lempel, A. (1977). "A universal algorithm for sequential data compression." IEEE Transactions on Information Theory, 23(3), 337-343. DOI
10. Alberts, B. (2014). "Molecular biology of the cell." 6th ed. New York, NY: Garland Publishing.
11. Li, M., et al. (2004). "The similarity metric." IEEE Transactions on Information Theory, 50(12), 3250–3264. DOI
12. Jiang, Z., et al. (2023). "Low-Resource Text Classification: A Parameter-Free Classification Method with Compressors." Findings of the Association for Computational Linguistics: ACL 2023, Toronto, Canada. Link
13. Juneja, S., et al. (2022). "An Approach to DNA Sequence Classification Through Machine Learning: DNA Sequencing, K Mer Counting, Thresholding, Sequence Analysis." International Journal of Reliable and Quality E-Healthcare, 11(2), 1–15. DOI
14. Orozco-Arias, S., et al. (2021). "K-mer-based machine learning method to classify LTR-retrotransposons in plant genomes." PeerJ, 9, e11456. DOI
15. Sarkar, B. K., et al. (2021). "Determination of k-mer density in a DNA sequence and subsequent cluster formation algorithm based on the application of electronic filter." Scientific Reports, 11(1). [DOI](http://dx.doi.org/10.
16. Ozan, S. (2023). "DNA Sequence Classification." GitHub Repository
17. Singh, N. (2023). "Demystify DNA Sequencing with Machine Learning." Kaggle Notebook
18. Singh, N. (2023). "DNA Sequence Dataset." Kaggle Dataset