Highly efficient and sensitive membrane-based concentration process allows quantification, surveillance, and sequencing of viruses in large volumes of wastewater.

Wastewater-based epidemiology is experiencing exponential development. Despite undeniable advantages compared to patient-centered approaches (cost, anonymity, survey of large populations without bias, detection of asymptomatic infected peoples…), major technical limitations persist. Among them is the low sensitivity of the current methods used for quantifying and sequencing viral genomes from wastewater. In situations of low viral circulation, during initial stages of viral emergences, or in areas experiencing heavy rains, the extremely low concentrations of viruses in wastewater may fall below the limit of detection of the current methods. The availability during crisis and the cost of the commercial kits, as well as the requirement of expensive materials such as high-speed centrifuge, can also present major blocks to the development of wastewater-based epidemiological survey, specifically in low-income countries. Thereby, highly sensitive, low cost and standardized methods are still needed, to increase the predictability of the viral emergences, to survey low-circulating viruses and to make the results from different labs comparable. Here, we outline and characterize new protocols for concentrating and quantifying SARS-CoV-2 from large volumes (500 mL-1 L) of untreated wastewater. In addition, we report that the methods are applicable for monitoring and sequencing. Our nucleic acid extraction technique (the routine C: 5 mL method) does not require sophisticated equipment such as automatons and is not reliant on commercial kits, making it readily available to a broader range of laboratories for routine epidemiological survey. Furthermore, we demonstrate the efficiency, the repeatability, and the high sensitivity of a new membrane-based concentration method (MBC: 500 mL method) for enveloped (SARS-CoV-2) and non-enveloped (F-specific RNA phages of genogroup II / FRNAPH GGII) viruses. We show that the MBC method allows the quantification and the monitoring of viruses in wastewater with a significantly improved sensitivity compared to the routine C method. In contexts of low viral circulation, we report quantifications of SARS-CoV-2 in wastewater at concentrations as low as 40 genome copies per liter. In highly diluted samples collected in wastewater treatment plants of French Guiana, we confirmed the accuracy of the MBC method compared to the estimations done with the routine C method. Finally, we demonstrate that both the routine C method processing 5 mL and the MBC method processing 500 mL of untreated wastewater are both compatible with SARS-CoV-2 sequencing. We show that the quality of the sequence is correlated with the concentration of the extracted viral genome. Of note, the quality of the sequences obtained with some MBC processed wastewater was improved by dilutions or enzyme substitutions suggesting the presence of specific enzyme inhibitors in some wastewater. To the best of our knowledge, our MBC method is one of the first efficient, sensitive, and repeatable method characterized for SARS-CoV-2 quantification and sequencing from large volumes of wastewater.

Cost of exome analysis in patients with intellectual disability: a micro-costing study in a French setting.

BACKGROUND: With the development of next generation sequencing technologies in France, exome sequencing (ES) has recently emerged as an opportunity to improve the diagnosis rate of patients presenting an intellectual disability (ID). To help French policy makers determine an adequate tariff for ES, we aimed to assess the unit cost per ES diagnostic test for ID from the preparation of the pre-analytical step until the report writing step and to identify its main cost drivers. METHODS: A micro-costing bottom-up approach was conducted for the year 2018 in a French setting as part of the DISSEQ study, a cost-effectiveness study funded by the Ministry of Health and performed in collaboration with the GAD (Génétique des Anomalies du Développement), a genetic team from the Dijon University Hospital, and a public sequencing platform, the Centre National de Recherche en Génomique Humaine (CNRGH). The analysis was conducted from the point of view of these two ES stakeholders. All of the resources (labor, equipment, disposables and reagents, reusable material) required to analyze blood samples were identified, collected and valued. Several sensitivity analyses were performed. RESULTS: The unit nominal cost per ES diagnostic test for ID was estimated to be €2,019.39. Labor represented 50.7% of the total cost. The analytical step (from the preparation of libraries to the analysis of sequences) represented 88% of the total cost. Sensitivity analyses suggested that a simultaneous price decrease of 20% for the capture kit and 50% for the sequencing support kit led to an estimation of €1,769 per ES diagnostic test for ID. CONCLUSION: This is the first estimation of ES cost to be done in the French setting of ID diagnosis. The estimation is especially influenced by the price of equipment kits, but more generally by the organization of the centers involved in the different steps of the analysis and the time period in which the study was conducted. This information can now be used to define an adequate tariff and assess the efficiency of ES. TRIAL REGISTRATION: ClinicalTrials.gov identifier NCT03287206 on September 19, 2017.

Doublecortin mutation leads to persistent defects in the Golgi apparatus and mitochondria in adult hippocampal pyramidal cells.

Human doublecortin (DCX) mutations are associated with severe brain malformations leading to aberrant neuron positioning (heterotopia), intellectual disability and epilepsy. DCX is a microtubule-associated protein which plays a key role during neurodevelopment in neuronal migration and differentiation. Dcx knockout (KO) mice show disorganized hippocampal pyramidal neurons. The CA2/CA3 pyramidal cell layer is present as two abnormal layers and disorganized CA3 KO pyramidal neurons are also more excitable than wild-type (WT) cells. To further identify abnormalities, we characterized Dcx KO hippocampal neurons at subcellular, molecular and ultrastructural levels. Severe defects were observed in mitochondria, affecting number and distribution. Also, the Golgi apparatus was visibly abnormal, increased in volume and abnormally organized. Transcriptome analyses from laser microdissected hippocampal tissue at postnatal day 60 (P60) highlighted organelle abnormalities. Ultrastructural studies of CA3 cells performed in P60 (young adult) and > 9 months (mature) tissue showed that organelle defects are persistent throughout life. Locomotor activity and fear memory of young and mature adults were also abnormal: Dcx KO mice consistently performed less well than WT littermates, with defects becoming more severe with age. Thus, we show that disruption of a neurodevelopmentally-regulated gene can lead to permanent organelle anomalies contributing to abnormal adult behavior.

Integrative and comparative genomic analyses identify clinically relevant pulmonary carcinoid groups and unveil the supra-carcinoids.

The worldwide incidence of pulmonary carcinoids is increasing, but little is known about their molecular characteristics. Through machine learning and multi-omics factor analysis, we compare and contrast the genomic profiles of 116 pulmonary carcinoids (including 35 atypical), 75 large-cell neuroendocrine carcinomas (LCNEC), and 66 small-cell lung cancers. Here we report that the integrative analyses on 257 lung neuroendocrine neoplasms stratify atypical carcinoids into two prognostic groups with a 10-year overall survival of 88% and 27%, respectively. We identify therapeutically relevant molecular groups of pulmonary carcinoids, suggesting DLL3 and the immune system as candidate therapeutic targets; we confirm the value of OTP expression levels for the prognosis and diagnosis of these diseases, and we unveil the group of supra-carcinoids. This group comprises samples with carcinoid-like morphology yet the molecular and clinical features of the deadly LCNEC, further supporting the previously proposed molecular link between the low- and high-grade lung neuroendocrine neoplasms.

Minor allele of the factor V K858R variant protects from venous thrombosis only in non-carriers of factor V Leiden mutation.

Factor V serves an important role in the regulation of blood coagulation. The rs6025 (R534Q) and rs4524 (K858R) polymorphisms in the F5 gene, are known to influence the risk of venous thrombosis. While the rare Q534 (factor V Leiden) allele is associated with an increased risk of venous thrombosis, the minor R858 allele is associated with a lower risk of disease. However, no study has deeply examined the cumulative impact of these two variations on venous thrombosis risk. We study the association of these polymorphisms with the risk of venous thrombosis in 4 French case-control populations comprising 3719 patients and 4086 controls. We demonstrate that the Q534 allele has a dominant effect over R858. Besides, we show that in individuals not carrying the Q534 allele, the protective effect of the R858 allele acts in a dominant mode. Thrombin generation-based normalized activated protein C sensitivity ratio was lower in the 858R/R homozygotes than in the 858K/K homozygotes (1.92 ± 1.61 vs 2.81 ± 1.57, p = 0.025). We demonstrate that the R858 allele of the F5 rs4524 variant protects from venous thrombosis only in non-carriers of the Q534 allele of the F5 rs6025. Its protective effect is mediated by reduced factor VIII levels and reduced activated protein C resistance.

Whole genome sequencing identifies a de novo 2.1 Mb balanced paracentric inversion disrupting FOXP1 and leading to severe intellectual disability.

The FOXP1 gene, located on chromosome 3p13, encodes the Forkhead-box protein P1, one of the four forkhead transcription factors which repress transcription by forming active homo- and heterodimers and regulate distinct patterns of gene expression crucial for embryogenesis and normal development. FOXP1 mutations, mostly truncating, have been described in patients with mild to moderate intellectual disability (ID), autism spectrum disorder (ASD), and speech and language impairment (MIM #613670). Here, we report a small de novo heterozygous balanced inversion of 2.1 Mb located at 3p14.1p13 identified by Whole Genomic Sequencing (WGS) and disrupting the genes FAM19A4 and FOXP1. This inversion was found in a patient with severe ID, ASD, seizures and very unusual vascular anomalies which were never described in the clinical spectrum of FOXP1 mutations. We show that the neurodevelopmental phenotype observed in the patient most likely results from FOXP1 haploinsufficiency as this heterozygous inversion leads to a 60 to 85% decrease of FOXP1 mRNA levels and to the complete absence of FOXP1 full-length protein. These findings, in addition to expanding the molecular spectrum of FOXP1 mutations, emphasize the emerging role of WGS in identifying small balanced chromosomal rearrangements responsible for neurodevelopmental disorders and not detected by conventional cytogenetics.

Novel mutations in DNAJB6 cause LGMD1D and distal myopathy in French families.

BACKGROUND AND PURPOSE: The aim was to determine the genetic background of unknown muscular dystrophy in five French families. METHODS: Twelve patients with limb girdle muscular dystrophy or distal myopathy were clinically evaluated. Gene mutations were identified using targeted exon sequencing and mutated DNAJB6 was tested in vitro. RESULTS: Five patients presented with distal lower limb weakness whilst others had proximal presentation with a variable rate of progression starting at the mean age of 38.5 years. Two novel mutations (c.284A>T, p.Asn95Ile, two families; and c.293_295delATG, p.Asp98del, one family) as well as the previously reported c.279C>G (p.Phe93Leu, two families) mutation in DNAJB6 were identified. All showed a reduced capacity to prevent protein aggregation. CONCLUSIONS: The mutational and phenotypical spectrum of DNAJB6-caused muscle disease is larger than previously reported, including also dysphagia. The originally reported c.279C>G (p.Phe93Leu) mutation is now identified in four different populations and appears to be a mutational hotspot. Our report confirms that some DNAJB6 mutations cause distal-onset myopathy and hence DNAJB6 defects should be considered broadly in dominant muscular dystrophy families.

Protein S Heerlen mutation heterozygosity is associated with venous thrombosis risk.

Hereditary Protein S (PS) deficiency is a rare coagulation disorder associated with an increased risk of venous thrombosis (VT). The PS Heerlen (PSH) mutation is a rare S501P mutation that was initially considered to be a neutral polymorphism. However, it has been later shown that PSH has a reduced half-life in vivo which may explain the association of PSH heterozygosity with mildly reduced levels of plasma free PS (FPS). Whether the risk of VT is increased in PSH carriers remains unknown. We analyzed the association of PSH (rs121918472 A/G) with VT in 4,173 VT patients and 5,970 healthy individuals from four independent case-control studies. Quantitative determination of FPS levels was performed in a subsample of 1257 VT patients. In the investigated populations, the AG genotype was associated with an increased VT risk of 6.57 [4.06-10.64] (p = 1.73 10-14). In VT patients in whom PS deficiency was excluded, plasma FPS levels were significantly lower in individuals with PSH when compared to those without [72 + 13 vs 91 + 21 UI/dL; p = 1.86 10-6, mean + SD for PSH carriers (n = 21) or controls (n = 1236) respectively]. We provide strong evidence that the rare PSH variant is associated with VT in unselected individuals.

17q21.31 duplication causes prominent tau-related dementia with increased MAPT expression.

To assess the role of rare copy number variations in Alzheimer's disease (AD), we conducted a case-control study using whole-exome sequencing data from 522 early-onset cases and 584 controls. The most recurrent rearrangement was a 17q21.31 microduplication, overlapping the CRHR1, MAPT, STH and KANSL1 genes that was found in four cases, including one de novo rearrangement, and was absent in controls. The increased MAPT gene dosage led to a 1.6-1.9-fold expression of the MAPT messenger RNA. Clinical signs, neuroimaging and cerebrospinal fluid biomarker profiles were consistent with an AD diagnosis in MAPT duplication carriers. However, amyloid positon emission tomography (PET) imaging, performed in three patients, was negative. Analysis of an additional case with neuropathological examination confirmed that the MAPT duplication causes a complex tauopathy, including prominent neurofibrillary tangle pathology in the medial temporal lobe without amyloid-ß deposits. 17q21.31 duplication is the genetic basis of a novel entity marked by prominent tauopathy, leading to early-onset dementia with an AD clinical phenotype. This entity could account for a proportion of probable AD cases with negative amyloid PET imaging recently identified in large clinical series.

Autosomal recessive truncating MAB21L1 mutation associated with a syndromic scrotal agenesis.

We report on a boy with a rare malformative association of scrotum agenesis, ophthalmological anomalies, cerebellar malformation, facial dysmorphism and global development delay. The reported patient was carrying a homozygous frameshift in MAB21L1 detected by whole-exome sequencing, considered as the most likely disease-causing variant. Mab21l1 knockout mice present a strikingly similar malformative association of ophthalmological malformations of the anterior chamber and preputial glands hypoplasia. We hypothesize that MAB21L1 haploinsufficiency cause a previously undescribed syndrome with scrotal agenesis, ophthalmological anomalies, facial dysmorphism and gross psychomotor delay as remarkable hallmarks. Four cases from the literature were reported with features suggestive of a similar and recognizable clinical entity. We hypothesize that MAB21L1 should be the culprit gene in these patients.

SORL1 rare variants: a major risk factor for familial early-onset Alzheimer's disease.

The SORL1 protein plays a protective role against the secretion of the amyloid ß peptide, a key event in the pathogeny of Alzheimer's disease. We assessed the impact of SORL1 rare variants in early-onset Alzheimer's disease (EOAD) in a case-control setting. We conducted a whole exome analysis among 484 French EOAD patients and 498 ethnically matched controls. After collapsing rare variants (minor allele frequency ≤1%), we detected an enrichment of disruptive and predicted damaging missense SORL1 variants in cases (odds radio (OR)=5.03, 95% confidence interval (CI)=(2.02-14.99), P=7.49.10(-5)). This enrichment was even stronger when restricting the analysis to the 205 cases with a positive family history (OR=8.86, 95% CI=(3.35-27.31), P=3.82.10(-7)). We conclude that predicted damaging rare SORL1 variants are a strong risk factor for EOAD and that the association signal is mainly driven by cases with positive family history.

De novo deleterious genetic variations target a biological network centered on Aß peptide in early-onset Alzheimer disease.

We hypothesized that de novo variants (DNV) might participate in the genetic determinism of sporadic early-onset Alzheimer disease (EOAD, onset before 65 years). We investigated 14 sporadic EOAD trios first by array-comparative genomic hybridization. Two patients carried a de novo copy number variation (CNV). We then performed whole-exome sequencing in the 12 remaining trios and identified 12 non-synonymous DNVs in six patients. The two de novo CNVs (an amyloid precursor protein (APP) duplication and a BACE2 intronic deletion) and 3/12 non-synonymous DNVs (in PSEN1, VPS35 and MARK4) targeted genes from a biological network centered on the Amyloid beta (Aß) peptide. We showed that this a priori-defined genetic network was significantly enriched in amino acid-altering DNV, compared with the rest of the exome. The causality of the APP de novo duplication (which is the first reported one) was obvious. In addition, we provided evidence of the functional impact of the following three non-synonymous DNVs targeting this network: the novel PSEN1 variant resulted in exon 9 skipping in patient's RNA, leading to a pathogenic missense at exons 8-10 junction; the VPS35 missense variant led to partial loss of retromer function, which may impact neuronal APP trafficking and Aß secretion; and the MARK4 multiple nucleotide variant resulted into increased Tau phosphorylation, which may trigger enhanced Aß-induced toxicity. Despite the difficulty to recruit Alzheimer disease (AD) trios owing to age structures of the pedigrees and the genetic heterogeneity of the disease, this strategy allowed us to highlight the role of de novo pathogenic events, the putative involvement of new genes in AD genetics and the key role of Aß network alteration in AD.

Variations in the APP gene promoter region and risk of Alzheimer disease.

We genotyped five polymorphisms, including two polymorphisms with known effects on transcriptional activity, in a large cohort of 427 Alzheimer disease (AD) cases and 472 control subjects. An association between rs463946 (-3102 G/C) and AD was found and was confirmed in a replication sample of a similar size. By contrast, analysis of three recently described rare mutations influencing APP transcription did not confirm their association with AD risk.

Association between the MnSOD Ala-9Val polymorphism and development of schizophrenia and abnormal involuntary movements in the Xhosa population.

Reactive oxygen species (ROS)-mediated damage has been hypothesized to play a role in the development and poor outcome of schizophrenia, as well as the development of neuroleptic-induced abnormal involuntary movements. Recently, the functional polymorphism (Ala-9Val) in the manganese superoxide dismutase (MnSOD) gene (part of the antioxidant defense mechanism) was found to be associated with schizophrenia in a Turkish population. This study was aimed at replicating this finding in a Xhosa population. In addition, the role of Ala-9Val in abnormal involuntary movement and tardive dyskinesia development in the Xhosa population was also investigated. The schizophrenic patient group (n=286) and a healthy control group (n=243) were genotyped for the Ala-9Val polymorphism using heteroduplex-single stranded conformational polymorphism (HEX-SSCP) analysis. No significant difference in genotype or allele frequency could be observed between the schizophrenia and control group (P=0.294 and P=0.528 respectively). In addition no association could be found between the polymorphism and symptom severity (SANS and SAPS). The Xhosa schizophrenia patient group with abnormal involuntary movements (n=54) and a subgroup with tardive dyskinesia (n=30) was found to significantly differ in Ala-9Val genotype frequency (P=0.008 and P=0.011 respectively) compared to the Xhosa schizophrenia patient group without abnormal involuntary movements (n=204). However, no significant difference was found for the allele frequencies (P=0.955 and P=0.161). Further, using ANCOVA no association was found between AIMS score and genotype in the group with abnormal involuntary movements (P=0.1234). However, in the patient group with tardive dyskinesia an association was observed between genotype and AIMS score (P=0.0365). These results do not support a major role of the MnSOD Ala-9Val polymorphism in the development of schizophrenia or symptom severity in the Xhosa population. Yet it seems to be involved in the development of abnormal involuntary movements and tardive dyskinesia and may even modulate the severity of tardive dyskinesia.

Suicide attempts in an African schizophrenia population: an assessment of demographic risk factors.

This study investigated demographic variables, including affected sibling pair status, as risk factors for suicidal behavior in schizophrenia patients of African (Xhosa) descent. Xhosa subjects with schizophrenia were interviewed with the Diagnostic Interview for Genetic Studies (DIGS) and then stratified into two groups: those with ( n = 90) and those without ( n = 364) a history of previous suicide attempts. Demographic parameters (including gender, age, and social circumstances, sib ship) were then compared across these groups. Demographic predictors of suicide included sib ship status ( p = 0.038; OR = 1.7) and age of onset of illness ( p = 0.008; OR = 2.5). On further analysis of suicide in siblings, only a minority of sib pairs was found to be concordant for a lifetime history of suicide attempts (3%). These findings raise the possibility that affected sib pair status may be protective in nature. Given the counter-intuitive nature of this finding, further work is needed to replicate it, and to explore possible underlying mechanisms.

A culture-bound syndrome 'amafufunyana' and a culture-specific event 'ukuthwasa': differentiated by a family history of schizophrenia and other psychiatric disorders.

BACKGROUND: 'Amafufunyana' and 'ukuthwasa' are two culture-specific descriptive terms used by Xhosa traditional healers to explain aberrant behavioral and psychological phenomena. Some overlap between these conditions and schizophrenia (DSM-IV) is apparent. The aim of this study was to determine the extent to which amafufunyana and ukuthwasa were used as cultural explanatory models by traditional healers for DSM-IV-defined schizophrenia and whether there were significant phenomenological differences in schizophrenia symptoms in patients with the diagnosis of amafufunyana rather than ukuthwasa. SAMPLING AND METHODS: Xhosa patients with schizophrenia underwent a structured clinical diagnostic interview (Diagnostic Interview for Genetic Studies). The use of traditional diagnostic and treatment methods was assessed by structured open-ended interviewer-rated questions. The sample was then stratified for the presence/absence of a past/current diagnosis of amafufunyana and/or ukuthwasa. The clinical parameters were compared across groups by means of the chi2 or Student t tests. RESULTS: 247 adult subjects participated in the study. 106 (53%) patients reported a previous diagnosis of amafufunyana, and 9 (4.5%) reported a diagnosis of ukuthwasa. A family history of schizophrenia (p = 0.004) or any psychiatric disorder (p = 0.008) was more common in the ukuthwasa group. Subjects with a primary diagnosis other than amafufunyana or ukuthwasa were more likely to be married (p = 0.004), to have a history of stressor(s) prior to illness onset (p = 0.026), to be from a rural environment (p = 0.007) or to have a history of cannabis abuse/dependency (p = 0.015). CONCLUSION: The culture-bound syndrome amafufunyana and the culture-specific phenomenon of ukuthwasa are both used to explain symptoms in patients with schizophrenia (DSM-IV). Identification of cases as amafufunyana and ukuthwasa may correlate with a distinction between familial and sporadic cases of schizophrenia. Whether the positive connotations associated with ukuthwasa, as opposed to the more negative connotations associated with amafufunyana, hold any implications for the treatment or prognosis of schizophrenia remains to be clarified.

Multicenter linkage study of schizophrenia loci on chromosome 22q.

The hypothesis of the existence of one or more schizophrenia susceptibility loci on chromosome 22q is supported by reports of genetic linkage and association, meta-analyses of linkage, and the observation of elevated risk for psychosis in people with velocardiofacial syndrome, caused by 22q11 microdeletions. We tested this hypothesis by evaluating 10 microsatellite markers spanning 22q in a multicenter sample of 779 pedigrees. We also incorporated age at onset and sex into the analysis as covariates. No significant evidence for linkage to schizophrenia or for linkage associated with earlier age at onset, gender, or heterogeneity across sites was observed. We interpret these findings to mean that the population-wide effects of putative 22q schizophrenia susceptibility loci are too weak to detect with linkage analysis even in large samples.

Examination of IMPA1 and IMPA2 genes in manic-depressive patients: association between IMPA2 promoter polymorphisms and bipolar disorder.

Manic-depressive (bipolar) illness is a serious psychiatric disorder with a strong genetic predisposition. The disorder is likely to be multifactorial and etiologically complex, and the causes of genetic susceptibility have been difficult to unveil. Lithium therapy is a widely used pharmacological treatment of manic-depressive illness, which both stabilizes the ongoing episodes and prevents relapses. A putative target of lithium treatment has been the inhibition of the myo-inositol monophosphatase (IMPase) enzyme, which dephosphorylates myo-inositol monophosphate in the phosphatidylinositol signaling system. Two genes encoding human IMPases have so far been isolated, namely myo-inositol monophosphatase 1 (IMPA1) on chromosome 8q21.13-21.3 and myo-inositol monophosphatase 2 (IMPA2) on chromosome 18p11.2. In the present study, we have scanned for DNA variants in the human IMPA1 and IMPA2 genes in a pilot sample of Norwegian manic-depressive patients, followed by examination of selected polymorphisms and haplotypes in a family-based bipolar sample of Palestinian Arab proband-parent trios. Intriguingly, two frequent single-nucleotide polymorphisms (-461C>T and -207T>C) in the IMPA2 promoter sequence and their corresponding haplotypes showed transmission disequilibrium in the Palestinian Arab trios. No association was found between the IMPA1 polymorphisms and bipolar disorder, neither with respect to disease susceptibility nor with variation in lithium treatment response. The association between manic-depressive illness and IMPA2 variants supports several reports on the linkage of bipolar disorder to chromosome 18p11.2, and sustains the possible role of IMPA2 as a susceptibility gene in bipolar disorder.

New parkin mutations and atypical phenotypes in families with autosomal recessive parkinsonism.

The frequency of parkin mutations was evaluated in 30 families of highly diverse geographic origin with early-onset autosomal recessive parkinsonism. Twelve different mutations, six of which were new, were found in 10 families from Europe and Brazil. Patients with parkin mutations had significantly longer disease duration than patients without the mutation but with similar severity of disease, suggesting a slower disease course. Two patients with parkin mutations had atypical clinical presentation at onset, with predominant tremor when standing.

Association studies in candidate genes: strategies to select SNPs to be tested.

OBJECTIVE: When numerous single nucleotide polymorphisms (SNPs) have been identified in a candidate gene, a relevant and still unanswered question is to determine how many and which of these SNPs should be optimally tested to detect an association with the disease. Testing them all is expensive and often unnecessary. Alleles at different SNPs may be associated in the population because of the existence of linkage disequilibrium, so that knowing the alleles carried at one SNP could provide exact or partial knowledge of alleles carried at a second SNP. We present here a method to select the most appropriate subset of SNPs in a candidate gene based on the pairwise linkage disequilibrium between the different SNPs. METHOD: The best subset is identified through power computations performed under different genetic models, assuming that one of the SNPs identified is the disease susceptibility variant. RESULTS: We applied the method on two data sets, an empirical study of the APOE gene region and a simulated study concerning one of the major genes (MG1) from the Genetic Analysis Workshop 12. For these two genes, the sets of SNPs selected were compared to the ones obtained using two other methods that need the reconstruction of multilocus haplotypes in order to identify haplotype-tag SNPs (htSNPs). We showed that with both data sets, our method performed better than the other selection methods.