SÍMBOLOS Y NOTACIÓN DE CARGAS

RFLPs (Restriction Fragment Length Polymorphisms) were introduced in

1978 (Kan & Dozy, 1978), (Botstein et aL, 1980) and became the first genetic

markers to be used in a successful linkage (Huntington’s disease) (Gusella et al.,

1983). They are based on a single base pair change that creates or obliterates a

cleavage site for a specific restriction enzyme. The resulting variation between

individuals can be detected by digestion of the DNA by the appropriate restriction

enzyme. RFLPs are inherited as simple Mendelian codominant markers, which can

be readily identified in families. A disadvantage of RFLPs is that because they are

biallelic they are not very informative, having low heterozygosity (usually<0.40).

The heterozygosity of the genotyping markers was greatly increased by the

identification of VNTRs (Variable Number of Tandem Repeats) in 1985 (Nakamura

et al., 1987). This class of marker is made of a specific set of consensus sequences

that vary between 14 and 100 base pairs in length. They are remarkably polymorphic,

with a high heterozygosity rate (usually > 0.6) in the population. However, there is

only a small number of VNTRs available and their distribution is rather limited in the

genome, often tending to cluster toward human telomeres. In addition, their high

heterozygosity together with their large size make these markers difficult to genotype.

STRs (Short Tandem Repeats) or microsatellites were initially described by

Weber and May (Weber & May, 1989) and Litt and Luty (Litt & Luty, 1989). STRs

are distributed widely and evenly in the genome. They sometimes have high

heterozygosities (usually > 0.7) and are relatively easy to score. The number o f the

(trinucleotide) or four (tetranucleotide) bases. The dinucleotide (CA)n-(GT)n is the

most common repeat, with a highly polymorphic form of the repeat occurring

approximately every 0.4 cM. Their polymorphic status usually depends on two

factors: the size of the repeat and whether a perfect or imperfect repeat is present. A

marker having 15, or more, perfect (uninterrupted) repeat motifs is usually, but not

always, highly polymorphic (Weber, 1990). Following the success of the dinucleotide

repeat markers, other polymorphic microsatellites were isolated and characterized,

namely, tri- and tetra-nucleotide repeat markers, which have helped to further

saturate, the genome with additional polymorphic markers for use in genetic studies.

Finally, in recent years, attention has been focused on the use of single

nucleotide polymorphisms (SNPs) as genetic markers. They are the most common

type of human DNA variation and as the name suggests they represent a position at

which two alternative bases occur at an appreciable frequency (>1%) in the human

population (Wang et al.^ 1998). On average, they occur 1 per 300-1000 base pairs

(Collins et al., 1997). Although individual SNPs are less informative than typical

multi-allelic simple sequence length polymorphisms, they are more abundant and

their genotyping can be automated with the use of DNA chip-based microarrays

(Hacia et al., 1996), which allow the very rapid analysis of very large numbers of

SNPs. Large collections of mapped SNPs are being developed and will provide a

1.7.2 Param etric approach: Led score analysis

The term genetic linkage is used to describe the tendency of alleles from two

or more loci, which lie in close physical proximity to each other to segregate together

in a family. Therefore, the extent of linkage depends on distance, which in genetic

terms can be measured with the frequency of recombination, also known as the

recombination fraction, or theta (0) between loci. Hence, the further apart the loci, the

greater the recombination fraction and vice versa. Theta ranges from 0 for loci that

are completely linked to 0.5 for unlinked loci on the same or different chromosomes.

Lod score analysis is a likelihood-based parametric linkage approach to

estimate the recombination fraction and the significance of the evidence for linkage.

Two likelihoods (L) of observing a particular configuration of a disease locus and a

genetic marker locus in a pedigree data set are calculated: one under various degrees

of linkage over a selected range of recombination fractions (i.e., 0 < 0.5) and the other

under no linkage (i.e., 0 = 0.5). The logio of the ratio for these two likelihoods, also

known as lod score z(x) is used as a measure of support for linkage versus non­

linkage (Morton, 1955):

z(x) = logio [ L(pedigree given 0=jc) / /.(pedigree given 0=0.5) ]

where jc is a particular value of theta. The value is termed the two-point lod when it

involves linkage between the disease locus and a genetic marker locus and multilocus,

or multipoint lod score when it involves linkage between a disease locus and several

linked marker loci. Lod scores across families can be summed but only at the same

recombination fraction and the overall lod score o f the data set is then examined.

Morton suggested that lods o f 3 or more are indicative of linkage because

theoretical considerations based on Bayesian probability and empirical evidence

positive. A lod score of -2 or less are indicative of non-linkage, while values between

3 and -2 are considered as inconclusive (Morton, 1955).

The lod score (or parametric) method of linkage analysis requires a genetic

model for the disease locus as well as specification of certain parameters such as:

recombination fraction, disease allele frequencies and penetrance values for each

possible disease genotype. Since a model of inheritance is unknown for

schizophrenia, researchers have tended to use a range of different models in which the

parameters are assumed. However, power to detect linkage is decreased when the

wrong assumptions are made. Several studies have shown that the problem with

applying a parametric method when the genetic model is unknown is not the false

conclusion of linkage but rather the increased probability of missing a linkage when it

is present (Clerget-Darpoux et a/., 1986).

One way to circumvent mis-specification of penetrance and/or the existence of

phenocopies is by performing an affecteds-only analysis in which one records

unaffected individuals as “phenotype unknown”. In this way maximum likelihood

estimates come from the most certain of the trait information, the phenotypes of

affected individuals. However, the inheritance pattern remains to be specified.

Another important issue when performing linkage analysis for complex traits

is locus heterogeneity. The lod score method provides statistical tests to calculate

heterogeneity (Ott, 1999). In this case, the likelihood ratio for linkage given

heterogeneity (hlod) may be written as

Z{ff) = logio [L{ 6, a)IL{6 =1/2, a = 0)]

where a the proportion of families being linked

The lod score method has several advantages when the mode of inheritance is

recombination fraction and it enables formal tests of genetic heterogeneity. However,

for complex genetic traits such as schizophrenia, with an unknown mode of

transmission it is necessary either to use arbitrary penetrances to evaluate linkage or

to use a “model free” method in which penetrance is varied automatically such that

the maximum lod is produced with gene frequency and penetrances that are

appropriate. Several investigators have argued for alternative approaches to the

detection of linkage described below.