Genetic analysis of complex traits

1
Genetic analysis of complex traits

• Studying inheritance of traits that show no clear
  Mendelian inheritance but cluster in families
• Suggests there is a genetic component
• Usually a mix of genetic and environmental
  factors
• Several models
• Single gene modified by environment
• Several genes each with significant contributions
  (oligogenes)
• Many genes each making a small contribution
  (polygenes)
• Last two- multifactorial inheritance

2
Problems of complex diseases

• Identification of involved genes hampered by
• genetic heterogeneity
• Alleles at more than one locus can trigger a
  specific disease
• reduced penetrance
• Individuals with predisposing genotype are
  unaffected
• phenocopy
• Disease is triggered by environmental factors in
  the absence of a predisposing genotype

3
Mapping using a parametric model

• In monogenic disorders, mapping was achieved
  through pedigree analysis in which linkage is
  sought between markers and the disease gene using
  LOD score analysis to combine data from multiple
  families
• In complex diseases, often there is a major locus
  that segregates in a Mendelian fashion (often
  with incomplete penetrance)
• In some families, the disease shows such an
  inheritance and can be mapped with LOD score
  analysis

4
LOD (Log ratio of odds) scores

• Used to overcome not having large numbers of
  progeny to accurately measure recombination
  frequency
• Tests a genetic model which states that the
  disease locus and a genetic marker are linked
  with a recombination fraction of ? and which
  requires that parameters are specified
• Dominant or recessive
• Degree of penetrance
• Allele frequencies
• Logarithm of the ratio of odds (Z) of the
  observed outcome if two loci are linked with a
  recombination fraction of ? to the odds of the
  observed outcome if they are unlinked (? 0.5)
• Zx odds of observed results if ? x odds
  of observed results if ? 0.5
• LOD ? x log 10 Zx

5
LOD scores

• Calculation can be repeated using different
  values of ? (ie. different distances between gene
  and marker)
• Result is a numerical value that measures the
  odds that two loci are linked at a given
  recombination fraction (ie. distance), compared
  with chances they are unlinked
• Threshold for declaring linkage is a LOD score of
  3 or more (ie. Z 10001) (translation- if there
  is a 10001 odds that the locus is linked with a
  certain distance (?) versus being unlinked)
• Unlikely that one family would give a LOD 3
  can combine data from a number of families
• MLS maximum likelihood score
• LOD score for the most likely of a series of
  alternatives
• Model with highest LOD score is judged most
  likely to be correct
• Since parameters have to be specified in these
  calculations, analysis is called parametric

6
example of LOD score analysis

• Are A and B linked?
• All children inherited aB from II-2
• Other chromosome came from II-1
• If ? 0 (totally linked), then for each child,
  chance of observed result is 0.5 (since it gets
  one of two chromosomes from II-1)
• For the 3 children, odds of observed results are
  (0.5)3 0.125
• If ? 0.5 (loci unlinked), then there is 0.25
  chance of the observed genotype (independent
  assortment) odds are (0.25)3 0.0156

7
LOD score analysis

• Zx odds of observed results if ? x odds
  of observed results if ? 0.5
• Z ? 0 0.125 8 0.0156
• LOD ? 0 log 10 Zx 0.9
• try again using a different ? 0.25 (linked with
  RF of 0.25)
• Now odds of observed results are
• 0.5 x 0.75 0.375? for family (0.375)3 0.053
• Why 0.75? Since odds of inheriting
  non-recombinant chromosome (as is seen in
  progeny) is 0.75 and there are two possible
  chromosomes that could be inherited
• Z ? 0.25 0.053 3.39 0.0156
• LOD ? 0.25 log 10 Zx 0.53

8
LOD score analysis

• So if we compare the two scenarios, with a ? 0,
  which means the loci are totally linked , we get
  a LOD score of 0.9
• With ? 0.25, which means the loci can be
  separated by recombination 25 of the time, we
  get a LOD score of 0.53
• Since ? 0 gives higher score, this is the MLS,
  meaning loci are most likely linked, although LOD
  lt 3 makes it not statistically significant
• Since LOD scores are logarithmic, they can be
  added, so 4 similar families would combine to
  give a score of 3.6 at ? 0.

9
Parametric models have mapped disease genes

• BRCA1
• Analyzed 23 families that showed clustering,
  indicating a familial mode of inheritance
• Linkage to a specific marker with a LOD 2.35
• When age of onset was considered
• analysis was restricted to 7 families with age of
  onset lt 45
• Gave LOD score of 5.98
• Alzheimer disease
• Hereditary non-polyposis colorectal cancer
• Familial adenomatous polyposis
• Non-insulin dependent diabetes

10
Psychiatric disorders

• Schizophrenia and manic depression show
  clustering but not Mendelian inheritance
• Attempts to map traits have not been successful
• Many reports that could not be replicated
• problems encountered
• Assessment of phenotype is variable
• Genetic heterogeneity
• LOD scores are very sensitive to status of a few
  key individuals in families- if their phenotype
  changes, affects outcome of analysis greatly

11
Multifactorial inheritance

• Those that involve two or more genes and a strong
  environmental influence
• Continuous variation
• Controlled by polygenic traits
• each involved gene contributes additively to
  phenotype
• Phenotypic expression of multifactorial traits
  varies widely due to gene interactions and
  environmental factors
• Polygenic traits
• Height
• Weight
• Skin colour

12
Multifactorial inheritance

• Other polygenic traits
• Neural tube defects
• Cleft palate
• Clubfoot
• Diabetes
• Hypertension
• Behavioural disorders
• Distributions of phenotypes in F2 varies
  depending on 2,3, or 4 genes involved
• As number of loci increases, the number of
  phenotypic classes increases, and the less
  phenotypic variation between classes
• Environmental influence smoothes out variation
  between genotypes even more

13
Model for inheritance of height

• Trait controlled by 3 genes with 2 alleles (A, a,
  B, b, C, c)
• Each dominant allele contributes equally to
  phenotype and recessive alleles make no
  contribution
• The effect of each dominant allele is additive
• Genes are not linked
• Assume
• base height of 5 feet
• Each dominant allele adds 3 inches
• aabbcc individual is 5 AABBCC individual is
  66
• If environment affects all people the same, the
  individuals with 3 dominant alleles (the most
  frequent genotype) would be the average height of
  59
• Genotype reflects genetic potential for height-
  environment (nutrition) will affect the full
  expression of the genotype

14
Environmental effects

• 3 genes give rise to 7 genotypic classes
• Environment blurs the distinction- actually see
  continuous variation

15
Liability

• take height example and apply to disease state
• If genetic risk is modified by environment-
  susceptibility (or liability) is normally
  distributed
• How many pre-disposing alleles are present? If
  three bi-allelic loci are involved, then 0 - 6
• Some traits do not show continuous variation in
  phenotypes ? either affected or not
• depends on threshold

16
Linkage and association

• Linkage studies use individual families where
  members are affected and attempt to demonstrate
  linkage between the occurrence of the disease and
  genetic markers (creates associations within
  families, but not among unrelated people)
• Association studies are based on populations and
  attempt to show an association between a
  particular allele and susceptibility to disease
  (a statistical statement about the co-occurrence
  of alleles or phenotypes)

17
Non-parametric linkage analysis

• parametric methods require a genetic model, and
  are only useful for complex traits with a single
  genetic component
• model-free or non-parametric linkage analysis
• ignores unaffected people
• look for alleles or chromosomal segments that are
  shared between affected people (within families
  or in whole populations association studies)
• if a gene contributes to the disease, then the
  genomic region will be co-inherited from a common
  ancestor by members of an affected pedigree more
  often than would be expected by chance

18
Genome scan

• follows the inheritance of of polymorphic
  micro-satellite markers in members of a pedigree
• if affected members co-inherit the same allele
  more often than expected by chance, then that
  genomic region may contain a gene that
  contributes to susceptibility
• usually based on affected sibling pairs
• involves genotyping markers evenly spread
  throughout genome
• sibs are expected to share 0, 1 or 2 alleles at
  an expected ratio of 25, 50 and 25

19
IBS and IBD

• need to distinguish DNA segments that are
  identical by descent (IBD) versus identical by
  state (IBS)
• IBS alleles look the same but are not derived
  from a common known ancestor
• IBD are copies of the same ancestral (usually
  parental) allele
• analysis most informative if there are multiple
  alleles or if there is a multi-locus multi-allele
  haplotype
• IBD studies require parental samples

20
IBS versus IBD
both share allele A1 IBS IBD only
obvious if parental genotype is known
21
sib pair analysis
Shares 2
Shares 1
Shares 1
Shares 0

• share 1 (1/4), 2 (1/2) or 0 (1/4) parental
  haplotypes by random segregation

• pairs of sibs affected by dominant condition
  share 1 or 2 haplotypes

• pairs of sibs affected by recessive condition
  share both parental haplotypes for affected region

22
Genome scanning- to identify commonly inherited
areas

• assemble families containing 2 affected sibs
• DNA samples from sibs and parents are collected
• microsatellites are genotyped using PCR with
  fluorescent tagged primers

23
Genome scanning cont

• Can analyze 18 loci in one lane and up to 500 on
  the same gel
• initial scan will cover whole genome in 10 cM
  intervals
• each locus will have two alleles distinguished by
  length of PCR product
• can determine parental and sib genotypes
• determine which alleles are IBD and how these
  vary from what is predicted based on allele
  frequencies
• big question- how do you decide whether excessive
  allele sharing is statistically significant?
• use LOD score analysis with lower MLS 1 as
  starting point
• try linkage disequilibrium studies

24
Association studies

• statistical statement about the co-occurrence of
  alleles or phenotypes
• allele A is associated with disease D if people
  who have D also have A more (or less) often than
  would be predicted from individual frequencies of
  A and D in the population
• eg. HLA-DR4 is found in 36 of UK population, but
  in 78 of people with rheumatoid arthritis
• population associations depend on population
  history
• in the UK, two unrelated people share a common
  ancestor 22 generations ago (500 years)(44
  meioses)
• if they inherit a disease susceptibility allele
  from their common ancestor, then during the many
  meioses, recombination will have reduced the
  shared segment to a small region. Only tightly
  linked loci will still be shared

25
Linkage disequilibrium

• non-random association in a population of
  alleles at two closely linked loci (so one allele
  closely linked to disease)
• based on having a common ancestor
• alleles that are closely linked will be commonly
  inherited
• but, in time, disequilibrium will disappear due
  to recombination (ie. Allele frequencies will
  equalize)
• if two alleles 1 Mb apart are in disequilibrium,
  then in 70 generations the disequilibrium will
  decay by 50
• more distantly located alleles will decay faster-
  get a gradient of disequilibrium, with highest
  value being closest to gene
• L.D. influenced by populations- effect is
  greatest in small, homogeneous populations
  (greatest chance of founder effects) (Finland),
  smallest in large heterogeneous populations (USA)

26
Other reasons for associations

• direct causation
• having allele A makes you susceptible to disease
  D (increases the likelihood)
• expect to see same allele A associated with
  disease in any population (bypasses common
  ancestor)
• natural selection
• people with disease may have a competitive
  advantage if they also carry allele A
• these are unlikely if the associated DNA is a
  variant in non-coding DNA
• studies in ethnically diverse populations are
  useful to distinguish between these causes and
  L.D.

27
Transmission disequilibrium test (TDT)

• tests done to check the results of an association
  study
• confirm whether a parent heterozygous for an
  associated and a non-associated allele transmits
  the associated allele more often to affected
  offspring
• starts with couples with more than one affected
  offspring
• select parent that is heterozygous for marker M1
• test compares number of such parents who transmit
  the M1 allele to their affected offspring versus
  transmitting other allele
• can be used if only one parent is available
• fundamentally a test of association, not linkage