Saturation of the human phenome

Abstract

The phenome is the complete set of phenotypes resulting from genetic variation in populations of an organism. Saturation of a phenome implies the identification and phenotypic description of mutations in all genes in an organism, potentially constrained to those encoding proteins. The human genome is believed to contain 20-25,000 protein coding genes, but only a small fraction of these have documented mutant phenotypes, thus the human phenome is far from complete. In model organisms, genetic saturation entails the identification of multiple mutant alleles of a gene or locus, allowing a consistent description of mutational phenotypes for that gene. Saturation of several model organisms has been attempted, usually by targeting annotated coding genes with insertional transposons (Drosophila melanogaster, Mus musculus) or by sequence directed deletion (Saccharomyces cerevisiae) or using libraries of antisense oligonucleotide probes injected directly into animals (Caenorhabditis elegans, Danio rerio). This paper reviews the general state of the human phenome, and discusses theoretical and practical considerations toward a saturation analysis in humans. Throughout, emphasis is placed on high penetrance genetic variation, of the kind typically asociated with monogenic versus complex traits.

Keywords: Human genome; genetics; phenome; saturation mutagenesis..

Fig. (1). Monogenic versus complex phenotypes.

Genetic variants may be anywhere between vanishingly rare (with a minor allele frequency at a minimum of one in the entire human population of approximately 6 billion, thus an allele frequency of 1/12x10⁹ chromosomes) up to 50% (after which the minor allele becomes the major allele). The physiological effect of a genetic variant may be individually very strong (high penetrance) or weak (low penetrance). Rare high penetrance alleles have historically been identified in families, and studied by genome-wide scanning with dense polymorphic anonymous markers, currently SNPs, followed by statistical linkage analysis and gene resequencing to identify causal variants in the family. Common low penetrance alleles are typically identified in large case/control cohorts, and studied by genome-wide SNP genotyping followed by simple statistical tests of differential allele frequency in the two classes, or by more sophisticated linkage disequilibrium (haplotype) mapping. Common high penetrance variants are unusual, since these would normally be either quickly fixed or eliminated from breeding populations. One circumstance under which such variants can be maintained is balanced selection, where there are opposing selections on heterozygotes versus homozygotes (malaria and sickle cell anemia being the best documented example). Rare variants of small functional effect are easily discovered through random resequencing but difficult to study mechanistically.

Fig. (2). Saturation kinetics and genetic dominance.

As suggested by Wright, if two doses of a wild type gene in a diploid organism provide saturating levels of physiological function (2x^R), then loss of one copy through mutation would still provide most functionality (1x^R). Such a situation would lead to recessive genetic behavior, whereby a heterozygote lof mutation would have little biological impact. Wright hypothesized that this would be the case for the majority of genes. In some cases, two wild type gene doses might provide function in the subsaturing range (2x^D), such that loss of one dose in heterozygous lof mutation carriers would have substantial biological effect (1x^D), leading to observed dominance. Dominance caused by unusual gain-of-function alleles is not an element of this model.

Fig. (3). Metabolic control theory and gene networks.

As suggested by Kacser and Burns [184], measures of enzymatic rates in purified in vitro systems do not necessarily reflect true in vivo situations. In reality, metabolic networks are complex, often with alternative routes from substrate inputs to catabolite outputs. In such cases, changing the rate of one specific step (indicated by arrow M), either via genetic mutation or through a targeted chemical antagonist or agonist (i.e. a drug) might lead to only modest, or unpredictable effects on total flux through the system and on the equilibrium levels of intermediate metabolites. This could explain both dominance as well as functional redundancy, whereby mutation of most genes even in homozygotes is not lethal, even though the genes are “required” in a mechanistic sense. Similar situations could be the case for developmental regulatory networks, where the flow of information rather than chemical metabolites is concerned.