The affected sib pair (ASP) method depends on the fact that when two siblings inherit alleles at a given locus, by chance alone they will share the same two allele 25% of the time, one of the same alleles 50% of the time, and share no alleles 25% of the time. If both siblings are afflicted with the same disease and there is linkage, ill siblings will share marker alleles more frequently. ASP tests analyzed the proportion of shared alleles and its statistical probability in a sample of affected sibling pairs. An alternative is the affected pedigree member (APM) method, which considers how often each pair of the affected relatives of any relationship share marker alleles versus the ‘by chance’expectation. However, AMP tests are less frequently used in schizophrenia linkages studies as compared with LOD score and ASP test.

In its simplest form, linkage analysis aims to identify a chromosome segment that is inherited from the parents by all of the affected and none of the unaffected members in the family. These often involve the use of a series of polymorphic markers to screen the entire genome in groups of families or pairs of affected siblings. The part of a chromosome that is transmitted with the disease phenotype down each generation is the likely location of a disease related gene. Further efforts to characterize the gene can then focus on that region. Typically, linkage analysis uses families with a high incidence of disorder over two or more generations, or large cohorts of pairs of siblings, both affected by illness. Parametric linkage methods typically involve multigenerational pedigrees and require estimation of the mode of inheritance, gene frequencies, phenocopy rate, and penetrance of the disease genes. Nonparametric techniques such as the affected sibling pair approach have attracted more interest in schizophrenia genetic studies because they are model-free and more suitable for studies of complex diseases.

Linkage analysis offers a more general approach for localizing predisposing genes. This strategy was recognized early in psychiatry. A hypothesis-free approach for mapping genes is required for linkage strategy. The first linkage studies in psychiatry, which was published in 1969 (5), focused on specific chromosome regions and used clinical genetic markers (monogenic disease with known localization). In subsequent years, clinical genetic markers were still used exclusively in this field (e.g. monogenic diseases such as red-green blindness or for the activity variability of certain enzymes). However, in the early 1980s, DNA markers (polymorphic sites with known position in the genome) were first applied to unravel the genetic basis of psychiatric disorders. Lately, informative marker systems spanning the whole genome; genome-wide scans are available (6).

Allelic association (or linkage disequilibrium) studies complement linkage and are another important method to locate genes. A well–known example of allelic association is the increased frequency of specific HLA antigens in several diseases. When a disease is transmitted from parent to offspring, the disease causing gene is inherited along with a segment of chromosome lying alongside. The pattern of DNA variants (polymorphism) within the segment will also be transmitted and all descendents who inherit the disease from a founding individual will inherit some of the surrounding haplotype. Markers very close to the gene will show allelic association with the disease in a population that includes the descendants of the disease founder. Since meiotic recombination at each generation steadily erodes the transmitted haplotype, only markers very close to a disease gene continue to show association over short genetic distances. The simplest design is to compare the frequency of alleles of a polymorphic marker in cases and control drawn from the same population. Association is particularly suited to the study of candidate genes considered on theoretical grounds to be implicated in disease process. The ideal situation for detecting association is when the polymorphism itself directly influences the function of a candidate gene.

Association studies represent a powerful method to detect genes of small effect. In case control association studies, the frequency of genetic variant is compared in a group of affected individuals versus a group of individuals unaffected with the phenotype of interest. This design is extremely powerful and sensitive to difference in ethnicity between groups, which may produce false positive (or negative) results (7). Family-based association studies using designs such as the Transmission Disequilibrium Test (TDT) compare the frequency of transmission of candidate alleles from heterozygous parents with the frequency of non-transmission of candidates alleles. Other variants of this approach include sibling-based paradigms involving comparison of candidate allele frequencies between affected individuals and their unaffected siblings. Family based approaches are generally less powerful than case-control designs but offer advantage of protecting against ethnic stratification (8).

The systematic hypothesis-free genome-wide strategy required the following: systems of positional DNA markers placed densely on the whole genome (initially restrictive fragment polymorphism, then repeat length polymorphism and now single nucleotide polymorphisms markers); and sample of genetically informative families each with more than one affected case (e.g. extended pedigree with multiple cases or pair of affected siblings). Unfortunately, it has been shown that this strategy did not guarantee quick success for complex disorders (in psychiatry as well as in other areas of medicine) as contrast with monogenic diseases (6)

The history of the search for genes contributing liability to schizophrenia is around a quarter of century old, but it is always dashed with nonreplication of the finding. This has been so despite consistent evidence from family, twin and adoption studies of an important genetic contribution; the heritability (or proportion of variance in liability explained by additive genetic effects) of schizophrenia is estimated to be approximately 80% (11). The reasons for the difficulty in finding genes include the complexity of the phenotype, heterogeneity and lack of biological marker. The mode of transmission is multifactorial where non-genetic determinants are also operating. As has been pointed earlier, schizophrenia does not conform to a classical Mendelian pattern of inheritance and it is now clear that most, perhaps all, cases involve the combined effects of many genes, each conferring a small increase in liability to the disorder; not due to single gene of major effects. As a consequence, a single gene does not seem to cause the disorder; thus no causal disease genes, only susceptibility genes are operating. Otherwise a consistently replicable linkage signal should have been detected. Advancement has also been hampered by the relatively small size of many studies. Not only are large sample needed to detect small effects, but even larger samples are needed to replicate positive findings

One way of trying to overcome this problem is to combine the data from multiple studies. Several groups that have formed collaborative studies; and funding agencies such as the National Institute of Health (USA) are requesting that DNA samples be made available to all qualified investigators. Combining data from families selected in differing ways, from heterogenous ethnic population, using different markers for their genome scans and different methods of analyzing is problematic. Badner and Gershon (12) adapted a method originally proposed by R.A Fisher in which P–value from multiple studies may be combined after correcting each value from the size of the region containing the minimum P-value. An alternative approach used by Levinson et. al (13) divided the genome into segments or ‘bins’and used a ranking method to assess the degree of support across studies for linkage within each segment. The two methods yielded overlapping but somewhat different results.

There are several weaknesses in the association studies. Even if an association between a variant of candidate gene and disease is found, it is difficult to interpret for the following reasons: First, association studies focus on one among a large number of candidate genes (or might even proceed genome-wide); this is usually a case of multiple testing, and appropriately adjusted levels of significant have to be applied. Criteria for handling this source of false-positive result remain to be agreed. Second, cases and control have to be comparable by population-genetic background. It is difficult to ensure that this criterion is met just by adopting careful sampling methods. Third, once an association is found, it is difficult to decide between two probabilities: (a) that the marker allele itself impacts on the risk for the disease, or (b) a genetic variant near to the marker allele is the real determinant and is in linkage disequilibrium with the disease allele. Only functional studies are suitable to resolve this ambiguity.

In dealing with complex disorder such as schizophrenia, linkage analysis also has inherent difficulties such as: (i) The magnitude of gene effect. Given the results of genome scans in psychiatric disorders, the susceptibility genes are likely to contribute to small or modest effects. So far, linkage analysis has been enormously successful in detecting causal or major gene effects, but not small effects; whereas association studies are substantially more powerful in detecting minor or modest gene effects. (ii) Strength of the magnitude of linkage signals across populations. Some linkage signals turned out to be replicable only in comparable genetic backgrounds, and not in other populations; for instance in schizophrenia, some linkage findings on 8p, 9q, 15q were replicable exclusively in African populations; whereas that on 10p has so far been replicable only among Caucasian populations (14). (iii) Sample size. Small effect sizes for genes with mild effects (e.g. odd ratio of about 1.5) required unrealistically large numbers of families with multiple affected members. Risch and Merikangas (15 ) calculated that, for an odd ratio of 1.5, the number of families required to detect the gene by linkage analysis is at least 18,000 depending on the model; the currently available maximal family sample sizes (of about 200) are at best to identify genes with an odd ratio of about 4. In schizophrenia, each of multiple genes contributes only modest effects; the contribution of each susceptible or vulnerable gene is believed to be limited to an odds ratio of less than 2.0 (14).

PRODH is another positional candidate for schizophrenia. It is a mitochondrial enzyme involved in transferring redox potential across the mitochodrial membrane, first identified by Liu et al (21) in 2002. It was found on a 1.5 Mb region on chromosome 22q11, a region previously implicated in the etiology of schizophrenia both by linkage studies and by work on velo-cardia facial syndrome (VCFS) (22). In this disorder the genetic defect is a microdeletion of the 22q11 regions, and sufferers have a 20-30% chance of a schizophrenia-like syndrome (23). Subsequent studies by other researchers gave conflicting results. Clearly, more work is needed to determine the role of this gene in schizophrenia.

Although numerous linkage and association studies have been carried out on this candidate gene, but until 2002 these failed to produce any conclusive results. Then a very large case-control study, including over 700 probands and 4000 controls, by Shifman et al (24) found a significant association between schizophrenia and COMT haplotype in Ashkenazi Jews. They suggested that the Val/Met polymorphism had only a moderate or no effect on schizophrenia risk and that more than one functional polymorphism within the COMT gene may be responsible. COMT codes for a gene product that inactivates catecholamines including dopamine. It does so by methylating m-hydroxy groups. Like PRODH, this gene is located in the 22q11 linkage/ VCFS microdeletion region, and is therefore a prime candidate gene to be involved in the risk of developing schizophrenia.

Attention focused on RGS-4 following a microarray study finding that the brains of schizophrenics showed decreased RGS-4 expression (25), and because of the location of the gene in a linkage region on chromosome 1q21-q22 of RGS-4 (26). The function of RGS proteins is to decrease the effect of G protein coupled receptor agonists. This could link with current theories on the etiology of schizophrenia relating to activity of dopamine, serotonin or metabo-tropic glutamate receptors.

Schizophrenia has not been convincingly associated with polymorphism in genes related to dopaminergic function, although meta-analyses have suggested a small but significant association for homozygosity at a polymorphism in DRD3 (3q13.3) (27). A modest but significant association between schizophrenia and a polymorphism in the serotonin2A receptor gene (HTR2A, 13q14-q21) was reported in a meta-analysis (28). There is a number of evidence that suggested the role for glutamatergic dysfunction in the pathogenesis or pharmacology of schizophrenia (29). Cholecystokinin (CCK, 3p22p21.3), which modulates dopaminergic neurotransmission has also been hypothesized to play a role in schizophrenia (30). In extensive review of the recent literature on molecular genetics of schizophrenia, Levinson (31) found that new genome scan project, seen in the light of previous scans, provide support for schizophrenia candidate regions on chromosomes 1q, 2q, 5q, 6p, 6q, 8p,10p,13q,15q, and 22q.

There were several reports of susceptible genes or markers of vulnerability in schizophrenia-related qualitative traits such as neuropsychological measures (32), neuro-imaging measures (33), neurological signs (34) and clinical dimensions (35)

With the recently completed the haplotype structure map (‘HapMap’), rapid decline in the cost of very high-throughput SNP genotyping and other emerging technologies, it is likely to find more susceptible genes and the functional significance of variations within them. The HapMap, developed by The International HapMap Project (36) will provide the common patterns of DNA sequence variations in the human genome, by characterizing sequence variations, their frequencies, and correlations between them, in DNA samples from populations with ancestry from parts of Africa, Asia and Europe. It will help investigators across the globe to discover the genetic factors that contribute to susceptibility to disease, to protection against illness and to drug response. The HapMap will provide an important shortcut to carry out candidate-gene, linkage-based and genome-wide association studies, transforming an unfeasible strategy to a practical one.