A new estimator of the discovery probability

Area of Science:

• Genomics
• Computational Biology
• Statistical Ecology

Background:

• Species sampling problems are critical in ecology and biology, influencing species richness evaluation and rare species estimation.
• Genomic applications face unique sampling challenges due to massive populations (genomic libraries) and limited sequencing.
• Existing methods struggle with the scale and complexity of genomic data, necessitating flexible approaches.

Purpose of the Study:

• To develop a Bayesian nonparametric approach for inferring species sampling issues in genomic contexts.
• To predict the discovery probability of new species (genes) from additional sample sizes in genomic libraries.
• To provide a method for quantifying the detection rate of rare genes and the coverage of abundant genes.

Main Methods:

• Utilized a Bayesian nonparametric framework for flexibility in handling large genomic datasets.
• Developed a novel estimator for the probability of detecting species with specific frequencies in an enlarged sample.
• Derived a closed-form expression for the estimator, allowing exact evaluation.

Main Results:

• The proposed estimator accurately predicts the discovery probability in genomic libraries.
• Quantified the rate of rare gene detection and the sample coverage of abundant genes as sample size increases.
• Demonstrated the method's efficacy using two expressed sequence tags (EST) datasets.

Conclusions:

• The Bayesian nonparametric approach offers a robust solution for species sampling problems in genomics.
• The derived estimator provides valuable insights into gene discovery and library characterization.
• This method aids in understanding rare gene detection and overall genomic library coverage.