baseLess: lightweight detection of sequences in raw MinION data

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Portable sequencing devices like MinION offer accessible nucleic acid analysis but require substantial computational resources for base calling.
• High computational costs limit the widespread deployment of portable sequencers, especially in remote areas or for large-scale applications.
• Targeted sequence detection offers a computationally efficient alternative to full base calling for specific applications.

Purpose of the Study:

• To introduce baseLess, a novel computational tool for direct target sequence detection from electrical signals generated by portable sequencers.
• To reduce the computational burden of nucleic acid analysis for applications such as infectious disease monitoring and species identification.
• To demonstrate the feasibility of running sophisticated sequence analysis on inexpensive, low-power hardware.

Main Methods:

• Development of baseLess, a tool employing an array of small neural networks for efficient detection of fixed-size subsequences.
• Direct analysis of electrical signals from sequencing data, bypassing traditional base calling.
• Validation on an inexpensive single-board computer to assess performance and resource requirements.

Main Results:

• BaseLess accurately identifies reads from three closely related fish species.
• The tool successfully classifies sequences within complex mixtures of 20 bacterial species.
• Demonstrated accurate performance on resource-constrained hardware, highlighting its potential for field deployment.

Conclusions:

• BaseLess significantly reduces computational requirements for targeted nucleic acid analysis.
• The tool enables portable, low-cost, and efficient sequence detection applications.
• BaseLess facilitates the deployment of advanced genomic analysis in resource-limited settings.