Decontaminating eukaryotic genome assemblies with machine learning.

Area of Science:

• Genomics
• Bioinformatics
• Machine Learning Applications in Biology

Background:

• High-throughput sequencing enables de novo genome assembly, but DNA extracts often contain contaminants.
• Existing eukaryotic genome decontamination methods rely on nucleotide similarity, risking loss of target sequences.
• Rigorous decontamination of eukaryotic assemblies remains a challenge.

Purpose of the Study:

• To introduce a novel machine learning application for rigorous sequence classification in de novo genome assemblies.
• To evaluate the performance of a decision tree method compared to existing decontamination protocols.

Main Methods:

• Application of a decision tree, a machine learning algorithm, for classifying de novo assembled sequences.
• Utilized the decision tree's ability to incorporate any measured feature without requiring pre-identified descriptors.
• Compared the decision tree's classification performance against established decontamination protocols.

Main Results:

• The decision tree method demonstrated superior performance in classifying sequences within eukaryotic de novo assemblies.
• The approach accurately identified both target and contaminant sequences.
• The decision tree method proved to be efficient and readily implementable.

Conclusions:

• Decision trees offer a more effective and accurate solution for decontaminating eukaryotic genome assemblies.
• This method is efficient, easy to implement, and robust in distinguishing target from contaminant sequences.
• Decision trees have broad potential for classifying sequences based on measured descriptors in biological datasets.