Bound2Learn

Bound2Learn classifies DNA-bound proteins from single-molecule and single-particle tracking data using machine learning to distinguish bound versus diffusive states and enable estimation of residence times.

Key Features:

• Machine learning classification: Applies machine learning to classify individual tracks from single-molecule/single-particle tracking as DNA-bound or non-bound.
• Tracking software compatibility: Leverages output from widely-used tracking software to process and analyze tracks.
• Residence time estimation: Uses classified tracks to estimate residence times, i.e., the duration proteins spend bound to DNA.
• Validation: Validated in silico using computational simulations and on live-cell single-molecule data from Escherichia coli and Saccharomyces cerevisiae.
• Slow-kinetics handling: Addresses challenges posed by slow kinetic processes and extended tracking durations that complicate identification of DNA-bound states.

Scientific Applications:

• Residence time measurement: Enables accurate measurement of protein residence times to assess stability of protein complexes.
• Protein–DNA interaction dynamics: Supports investigation of DNA-binding dynamics and mechanisms of DNA-bound proteins in live cells.
• Cross-species single-molecule studies: Applicable to single-molecule data from model organisms including Escherichia coli and Saccharomyces cerevisiae for comparative analyses.
• Characterization of slow kinetics: Facilitates analysis of processes with slow kinetics in single-particle tracking experiments.

Methodology:

Applies machine learning to classify tracks derived from outputs of tracking software, with validation performed by in silico simulations and live-cell single-molecule tracking data from Escherichia coli and Saccharomyces cerevisiae.