Related Experiment Video
Updated: Jan 15, 2026

Using Phylogenetic Analysis to Investigate Eukaryotic Gene Origin
Published on: August 14, 2018
Machine Learning Enables Alignment-Free Distance Calculation and Phylogenetic Placement Using k-Mer Frequencies
Eleonora Rachtman1, Yueyu Jiang1, Siavash Mirarab1
1Department of Electrical and Computer Engineering, UC San Diego, San Diego, California, USA.
We introduce kf2vec, a novel method for phylogenetic placement of long DNA sequences using k-mer frequencies and deep learning. This approach simplifies analyses by avoiding sequence alignment and accurately identifies taxonomic labels for new genomic samples.
Area of Science:
- Genomics
- Bioinformatics
- Computational Biology
Background:
- Phylogenetics is crucial for ecological studies, particularly for assigning taxonomic labels to unknown sequences.
- Phylogenetic placement methods are well-developed for short DNA fragments but less so for longer sequences like genomes or contigs.
- Placing long sequences offers more phylogenetic signal but presents challenges in homology detection and computational load.
Purpose of the Study:
- To develop and evaluate a novel phylogenetic placement method for long DNA sequences.
- To address the limitations of existing methods in handling large genomic data and computational complexity.
- To improve the accuracy and efficiency of taxonomic identification for long DNA sequences.
Main Methods:
- A new method, kf2vec, utilizes k-mer frequencies to measure distances between long query sequences and reference genomes.
- kf2vec employs a deep neural network trained to estimate phylogenetic distances from k-mer frequency vectors, bypassing the need for sequence alignment.
- The method is applicable to any genomic region and does not require marker genes, simplifying bioinformatics pipelines.
Main Results:
- kf2vec demonstrates superior performance compared to existing k-mer-based approaches in distance calculation.
- The method achieves accurate phylogenetic placement and taxonomic identification of new samples, including assembled genomes, contigs, and long reads.
- kf2vec effectively handles the challenges associated with analyzing longer DNA sequences.
Conclusions:
- kf2vec offers a robust and efficient solution for the phylogenetic placement of long DNA sequences.
- The alignment-free nature and deep learning approach of kf2vec simplify genomic analysis and enhance taxonomic identification accuracy.
- This method advances the application of phylogenetics in ecological and genomic studies involving large sequence data.
Related Concept Videos
Evolutionary Relationships through Genome Comparisons
Modern Molecular Taxonomy
Applications of Molecular Taxonomy
Gene Evolution - Fast or Slow?
In contrast, regions which code...
Phylogenetic Trees
Phylogeny

