perbase

Perbase, often written as per-base or perbase, is a term used in genomics and bioinformatics to denote measurements, analyses, or summaries made at individual nucleotide positions across a reference genome. It is not a single standardized technology, but a descriptive label used in many pipelines to report metrics on a base-by-base basis.

Common per-base metrics include per-base quality scores, per-base coverage, and per-base error rates. Per-base quality scores reflect the probability that a given base call is incorrect at each position, typically represented by Phred scores in sequencing data. Per-base coverage indicates how many sequencing reads align to each base, informing confidence in detected variants. Per-base error rates summarize the frequency of incorrect base calls at each position across reads. These per-base measurements are fundamental to quality control, helping identify regions of low coverage, systematic sequencing errors, or biases related to technology, library preparation, or alignment.

In downstream analyses, per-base information serves as the foundation for higher-level summaries and decisions, such as variant calling, where base-level evidence is aggregated into genotype likelihoods and allele balance. Limitations of per-base analyses include susceptibility to noise from uneven coverage, repetitive regions, GC bias, and mapping errors. As a result, per-base results are often supplemented with smoothing, binning, or context-aware models to support robust interpretation.