2.9.1 Methodologies for identifying genes and basic genetic concepts

Chromosomes and alleles

• A normal diploid somatic cell holds 46 chromosomes in 23 homologous pairs: 22 pairs of autosomes (44 chromosomes) and one pair of sex chromosomes.

• A haploid gamete holds 23 chromosomes.

• A locus is a defined position in the genome, and an allele is one of the sequence versions found at a locus.

• Homologous chromosomes are the maternal and paternal copies of a chromosome; they share loci but may carry different alleles.

• A chromatid is one of the two identical copies of a chromosome made by DNA replication in S phase, held to its sister at the centromere until anaphase.

• Sister chromatids are identical copies made by replication, so they carry the same allele.

• Replication does not change the chromosome count, because chromosomes are counted by centromere: a metaphase cell has 46 chromosomes and 92 chromatids.

Tap to enlarge

Chromosomes are classed by where the centromere sits.

The cell cycle and mitosis

The cell cycle runs G1, S, G2, mitosis, then G1 again in each daughter cell.

• DNA replicates in S phase of interphase, before mitosis, not during division itself.

• Mitosis runs in one fixed order, set out below, and yields two genetically identical diploid cells.

Tap to enlarge

Meiosis and recombination

• Meiosis takes one diploid cell through one round of replication and two divisions to give four genetically distinct haploid cells.

• The first meiotic division separates homologues and the second separates sister chromatids.

• Meiosis, unlike mitosis, pairs homologous chromosomes and recombines them at chiasmata.

• Crossing over between non-sister chromatids of paired homologues happens in prophase I.

• Mendel's law of segregation states that the two alleles a parent carries for a gene separate when gametes form, so each gamete receives one.

• Independent assortment mixes the parental chromosome pairs into the gametes.

Tap to enlarge

X inactivation, mosaics and chimeras

• Lyonisation is the random inactivation of one X chromosome in each female somatic cell, and the inactive X forms the Barr body.

• Every X beyond the first is inactivated, so the number of Barr bodies is the X count minus one, whatever the person's sex.

• Lyonisation is not genomic imprinting, because which X is silenced is random rather than set by parent of origin.

• A mosaic has cell lines from one zygote, arising through a postzygotic (mitotic) change.

• A chimera has genetically distinct cell lines from more than one zygote.

DNA and RNA

• Deoxyribonucleic acid (DNA) has two antiparallel strands, and histone packaging controls how accessible it is.

• A nucleotide is a pentose sugar, a phosphate group and one nitrogenous base.

• A nucleoside is the sugar joined to a base with no phosphate; adding phosphate makes a nucleotide.

• DNA's sugar is deoxyribose, whereas the ribose of ribonucleic acid (RNA) carries an extra hydroxyl group on its 2' carbon.

• Purines are nitrogen-containing heterocycles with a fused double ring; pyrimidines have a single ring.

• RNA molecules are much shorter than DNA because each usually copies only a small part of the genome, such as one gene.

The genome and its regulation

• Only about 1 to 2 per cent of the human genome codes for protein.

• Non-coding variation in regulatory, RNA, intronic and repeat sequence can still affect phenotype.

• Promoters, enhancers and silencers regulate expression by cell type and developmental stage.

• The promoter is where transcription factors and RNA polymerase II assemble, so a variant that changes a gene's transcriptional activity most likely lies there.

• The transcriptome is the whole collection of RNA transcripts in a cell or tissue at the moment it is sampled.

Transcription and splicing

• Transcription is the synthesis of RNA from a DNA template by RNA polymerase, in the nucleus, and is the first step of gene expression.

• Transcription makes RNA, whereas replication makes DNA.

• Capping, polyadenylation and splicing turn precursor messenger RNA (pre-mRNA) into mature messenger RNA (mRNA).

• An intron is transcribed into pre-mRNA but removed by splicing, so it is absent from mature mRNA; exons remain.

• Splicing removes the introns and joins the exons, and the spliceosome carries it out.

• Alternative splicing lets one gene give several proteins (isoforms) by joining its exons in different combinations.

• Mature mRNA carries the coding sequence from the nucleus to the ribosome, where its codons set the amino acid order.

Tap to enlarge

The genetic code

• A codon is a triplet of mRNA bases that specifies one amino acid or a stop signal.

• The number of possible codons is the number of bases raised to the codon length: four bases in triplets give 64 codons, coding 20 amino acids and three stops.

• AUG is the start codon and codes for methionine; it is the only methionine codon.

• The three stop codons are UAA, UAG and UGA in mRNA (TAA, TAG and TGA in DNA).

• The code is degenerate: most amino acids have several codons, which often differ only at the third base.

Tap to enlarge

Translation

• Translation is the decoding of mRNA codons into protein on ribosomes in the cytoplasm, using transfer RNA.

• Normal mRNA levels with reduced protein point to a block at translation.

• Aminoacyl-tRNA synthetase charges each tRNA with its matching amino acid, using adenosine triphosphate (ATP).

After translation

• Phosphorylation is the transfer by kinases of a phosphate from ATP to serine, threonine or tyrosine residues of an existing protein; excess phosphorylation of tau leads to tangles.

• N-linked glycosylation adds sugar chains to asparagine in the endoplasmic reticulum and Golgi, which supports folding and delivery.

• Ubiquitin is the small protein tag attached to unwanted proteins, which the proteasome then degrades.

• The lysosome handles aggregates and organelles by autophagy.

• Failure of the ubiquitin-proteasome route is linked to neurodegeneration.

Mutations

• A frameshift is told from an in-frame change by whether the number of bases inserted or deleted divides by three.

• The commonest mutation type in Duchenne muscular dystrophy is a frameshift deletion.

• A transition swaps one pyrimidine for the other (cytosine and thymine) or one purine for the other (adenine and guanine).

Polymorphisms and copy number variants

• A polymorphism is a common variant, with the rarer allele at 1% or more, held by evolutionary processes and with no overt phenotypic effect.

• A mutation is rarer and may be pathogenic; a common missense change (above 1%) counts as a polymorphism.

Genotype, phenotype and gene action

• Genotype is the alleles a person carries, whether expressed or not.

• Phenotype is the observable result of genotype acting within an environment and through development.

• Penetrance is the proportion of carriers of a disease genotype who express the phenotype, so below 100% a carrier can be unaffected.

• Variable expressivity is variation in severity or range of features among carriers who all show the condition.

Incomplete penetrance means some carriers are unaffected; variable expressivity means affected carriers differ in how badly.

• Epistasis is interaction between genes at different loci, where one locus modifies or masks the effect of another.

• Epigenetic modification, such as DNA methylation, changes gene expression and behaviour with no change in base sequence.

• Genomic imprinting is parent-of-origin-specific expression with no change in DNA sequence, as at 15q11-q13 in Angelman and Prader-Willi syndromes.

• Locus heterogeneity is one disorder arising from variants in different genes.

• Phenotypic heterogeneity is one variant giving different disorders.

Mendelian inheritance

• Male-to-male transmission on a pedigree excludes X-linked and mitochondrial inheritance.

Tap to enlarge

Autosomal recessive risks in each pregnancy

• Two carriers therefore have a 3 in 4 chance of an unaffected child in each pregnancy.

Hardy-Weinberg and population genetics

• Allele frequency is the number of copies of the allele divided by 2N, where N is the number of people: each homozygote contributes two copies and each heterozygote one.

• Genotype frequency is counted over the N individuals, and the proportion carrying a given allele is heterozygotes plus homozygotes for that allele, over N.

• For two alleles with frequencies p and q, Hardy-Weinberg genotype frequencies are p squared, 2pq and q squared, summing to 1.

• 2pq is the expected heterozygote frequency, and observed heterozygotes below it show a deficit.

• Under Hardy-Weinberg equilibrium allele and genotype frequencies stay constant across generations.

• Equilibrium needs random mating, a large population and no mutation, migration or selection.

• A surplus of homozygotes with a matching shortfall of heterozygotes, once drift, gene flow, selection and technical error are excluded, points to non-random mating.

• Assortative mating is non-random choice of partners who resemble oneself in a phenotype, including psychiatric illness and substance use.

• The carrier frequency of an autosomal recessive disorder is 2pq, about 2q, where q is the square root of the incidence; stopping at q omits the factor of two.

• For an X-linked recessive trait, male prevalence is q and female prevalence q squared, because men are hemizygous.

Family, twin and adoption studies

• Second-degree relatives, such as half-siblings, share on average one-quarter of their genes by descent; first cousins share about one-eighth.

• Sibling relative risk (lambda s) is the sibling's risk divided by the population risk, a measure of familial clustering; it is about 10 in autism spectrum disorder.

Twins

• MZ twins share nearly all nuclear sequence, whereas DZ twins share about half of segregating variation.

• Most MZ pairs are monochorionic and share more of the intrauterine environment than DZ pairs, which are almost always dichorionic.

• Concordance means both twins share a named trait and discordance means only one does; both are trait-specific and do not reveal zygosity.

In the formulas below, C is the number of concordant pairs and D the number of discordant pairs.

• A large MZ excess over DZ concordance shows genetic influence, and MZ concordance well below 100% shows that unique environment also contributes.

• The equal environment assumption concerns trait-relevant similarity; differential treatment, intrauterine sharing and assortative mating can bias twin estimates.

Adoption

• Danish adoption studies showed that schizophrenia risk travels with biological, not adoptive, kinship, arguing against rearing alone as a cause.

• Selective placement and restricted, generally advantaged adoptive homes limit what adoption studies can show.

Heritability

• Heritability is the proportion of phenotypic variance in a population attributable to genetic variance.

• It is not an individual's risk, the probability of transmission, or the share of any person's disorder caused by genes.

• It shifts when environmental variance shifts, so it holds only for a given population and environment.

• Falconer's formula estimates heritability from twin data as twice the difference between the MZ and DZ correlations.

• Taken as a group, schizophrenia, bipolar disorder and ADHD sit at about 60 to 85%.

• Schizophrenia's 80% is a population variance figure and differs from its 40 to 50% MZ concordance.

• Common disorders such as schizophrenia and ADHD have a polygenic architecture of many common variants, each of small effect, alongside rarer variants of larger effect.

• Autism has heritability similar to ADHD, but more of its liability lies in rare variants and CNVs, which genome-wide association studies do not capture.

• In autism, MZ concordance is roughly 60 to 65% and same-sex DZ concordance for the broad spectrum roughly 30 to 35%.

• An affected first-degree relative raises the risk of alcohol dependence about fourfold.

• First-degree relatives of people with bipolar disorder have a two to threefold raised rate of unipolar depression.

Familial risk of schizophrenia

Gottesman's pooled data show lifetime risk falling stepwise with falling relatedness, and staying above the population rate even in third-degree relatives.

• The DZ co-twin risk is nearly double the sibling figure, probably because of the shared womb and early years.

• The MZ co-twin risk falls well short of 100%, which excludes genetic determinism.

• A 22q11.2 deletion carries about a one-in-four lifetime risk of schizophrenia, regardless of family links.

Endophenotypes

• An endophenotype is a heritable, state-independent intermediate trait between genotype and clinical illness, introduced by Gottesman and Shields.

• An endophenotype indexes inherited liability, whereas a biomarker may merely be associated with the disorder.

Gottesman and Gould set five criteria.

Finding genes

• Gene mapping locates a risk gene on a chromosome by linkage (pedigrees, sib pairs) or association (case-control or family-based designs).

• Segregation analysis fits dominant, recessive and multifactorial transmission models to pedigree data and comes before localising a gene.

• Direct analysis tests a variant itself, whereas gene tracking follows a linked family marker when the causal variant is not tested.

• A transgenic animal study moves a gene from one species into another to test its function, turning a statistical signal into a mechanism.

Linkage analysis

• Linkage is the co-inheritance of genes that lie near each other on a chromosome, because crossovers between them are rare.

• A haplotype is a set of alleles at closely linked loci on one chromosome inherited together as a block.

• Linkage analysis follows polymorphic markers through pedigrees to localise a gene by co-segregation with the disorder.

• It suits rare, highly penetrant, large-effect disorders in large families, not polygenic disorders.

• A genome-wide linkage scan types evenly spaced polymorphic markers in affected families with no candidate gene in mind.

Recombination and distance

• The recombination fraction is recombinant meioses divided by informative meioses, from 0 (complete linkage) to 0.5 (independent assortment).

• A fraction of 0.5 means distant loci or loci on different chromosomes; 2 recombinants in 40 meioses suggests proximity, and 20 in 40 gives 0.5.

• The marker with the lowest recombination fraction is the most tightly linked, however few recombinants another marker shows in absolute terms.

• The centimorgan is a unit of genetic distance equal to a 1% chance of recombination per meiosis, about a megabase of human DNA on average.

• Genetic mapping orders loci by recombination frequency measured in families, in centimorgans; physical mapping measures distance in base pairs from the DNA itself.

The LOD score

• The logarithm of the odds (LOD) score is the base-10 logarithm of the likelihood ratio for linkage at a given recombination fraction against no linkage.

• It tests for linkage and estimates the recombination fraction, and scores add across families.

Tap to enlarge

Association studies

• Association studies compare allele frequencies in cases against unrelated controls (case-control) or controls from within families.

• Family-based designs use trios analysed by the transmission disequilibrium test, which is still association, not linkage.

• Linkage disequilibrium is the non-random association of alleles at nearby loci across a population, which lets a non-functional marker flag a signal.

Linkage follows co-inheritance through pedigrees, whereas association compares allele frequencies between groups.

• A genome-wide association study (GWAS) compares allele frequencies of common variants between unrelated cases and controls across the genome, with no prior hypothesis.

• Because it runs about a million tests, genome-wide significance is usually set at p below 5 × 10^-8, with independent replication required.

• In a Manhattan plot, each point represents one SNP.

• A genome-wide significant signal shows association, not causation, because a lead non-coding SNP may tag a causal variant in linkage disequilibrium.

• Ancestry differences can confound associations; matching and principal components help.

• Polygenic scores sum weighted risk alleles to estimate relative liability within a reference population, and ancestry limits how well they transfer.

• Rare sequence variants need methods beyond ordinary common-variant tests.

Tap to enlarge

Genes and environment

• Gene-environment interaction means an exposure's effect depends on genotype, or the reverse.

• Gene-environment correlation concerns exposure itself: passive (parents provide genes and environment), evocative (others respond to a person's traits) or active (people select environments).

Candidate and disease genes