Improved detection of differentially represented DNA barcodes for high-throughput clonal phenomics

Area of Science:

• Genomics
• Computational Biology
• Cancer Research

Background:

• Cellular DNA barcoding is key for studying cell population heterogeneity and identifying clones with varying responses.
• Current methods for statistical inference of differentially responding clones lack reliability.
• A need exists for robust algorithms to analyze DNA barcode read count data accurately.

Purpose of the Study:

• To develop and validate an improved algorithm for statistical inference in DNA barcoding experiments.
• To create a realistic benchmark dataset for modeling clone-tracing experiments.
• To enhance the detection of differentially responding clones and deconvolute clonal subpopulations.

Main Methods:

• Generated a benchmark read count dataset using mixtures of DNA-barcoded cell pools.
• Developed an algorithm accounting for intrinsic statistical properties of DNA barcode read count data.
• Applied multidimensional phenotypic profiling to analyze clonal subpopulations.

Main Results:

• The improved algorithm demonstrated a significantly lower false-positive rate compared to RNA-seq analysis algorithms.
• The algorithm showed particular efficacy in detecting differentially responding clones under strong selection pressure.
• Demonstrated successful deconvolution of phenotypically distinct clonal subpopulations using phenotypic profiling.

Conclusions:

• The developed algorithm provides a more reliable statistical method for clone-tracing experiments.
• The benchmark dataset and analysis methodology offer a foundation for future algorithm development and validation.
• This work advances the ability to study cellular heterogeneity and identify therapeutically relevant clones.