PLI

pLI, short for probability of loss-of-function intolerance, is a gene-level metric derived from large-scale human population sequencing data (notably ExAC and gnomAD). It estimates how intolerant a gene is to loss-of-function (LoF) variants that truncate or disrupt the protein, such as stop-gain, frameshift, and essential splice-site changes. pLI represents the posterior probability that a given gene is extremely unlikely to tolerate heterozygous LoF variation.

The value is computed by comparing the observed number of pLoF variants in a gene to the number expected under a neutral model, using a Bayesian framework that assigns genes into probabilistic categories of intolerance. In practice, genes with pLI near 1 are considered LoF-intolerant, while genes with pLI near 0 are considered LoF-tolerant. A high pLI suggests a gene may be haploinsufficient or otherwise dosage-sensitive, and such genes are often enriched for dominant disease genes.

Uses of pLI include prioritizing candidate genes in rare disease studies, guiding interpretation of exome and genome sequencing results, and informing functional hypotheses about gene dosage sensitivity. pLI is frequently cited alongside other intolerance metrics and population-based constraints, such as LOEUF, which provides a robust upper-bound estimate of observed/expected LoF depletion.

Limitations should be noted: pLI depends on annotation accuracy, sequencing coverage, and sample representation, and it does not measure essentiality or causality for a specific disease or variant. It cannot distinguish mechanisms or predict penetrance, and it is less informative for non-protein-coding regions. As data mature, newer metrics may complement or refine the inferences drawn from pLI.