The SHGP is one of an array of national genome projects that have aimed to describe the genetic background of specific populations (Figure 2). The projects share common aims, namely, (1) the development of infrastructure to create a network of state-of-the-art human genome sequencing centers; (2) training to create the skills, capabilities and capacity for human genome sequencing; and (3) the ability to solve disease genetics and identify traits or disorders within a specific population that can facilitate the genetic distribution of disease within a country. The projects also share common drawbacks, including the incorrect interpretation of a pathogenic variant, resulting from an incomplete human reference genome; an incomplete exon being represented in public databases; and an incorrect annotation of transcripts/exons owing to differing tissue-specific expression patterns (14). The use of genetics to guide therapy is also not without critique. Genomic medicine, through the ability to control the basic genes of a human, may lead to a loss of human diversity, the development of “designer” humans, and a genetic racism for inherited diseases.

An array of national genome projects based on NGS approaches is well established, including projects in Iceland (deCODE), the United Kingdom, the Netherlands, Finland and Sweden (reviewed in (15)). The scale of interest in national genomic screening is highlighted by the emergence of several new sequencing programs, including the initiative on rare and undiagnosed diseases in Japan that was launched in 2015 through collaborations with more than 400 hospitals, including 34 clinical centers; the 100,000 genomes project in China that launched a five-year project in December 2017 and will sequence the genomes of 100,000 individuals from different Chinese ethnic backgrounds; the genome program in Qatar that aims to establish a Qatari reference genome map through the sequencing of 3,000 genomes (~1% of the Qatari population); the genomic health futures mission in Australia that launched in May 2018; the “all of us” research program in the United States that aims to compile data from ≥ 1 million individuals beginning with its inception in May 2018; the personalized medicine program in Estonia that launched in 2016 from an initial cohort of 51,535 gene donors; the genomic program (génomique) in France that was launched in 2016 that will encompass a network of 12 genomic sequencing centers; the Dubai genomics program in the UAE that is currently focusing on building genomic medicine infrastructure and initiating large-scale whole-genome sequencing; and the Turkish genome project in Turkey that plans to sequence 1 million genomes by 2023. When compared to the scale of these studies, the SA population of ~32 million may appear low, but the SHGP has developed ≥13 gene panels covering ≥ 5,000 inherited diseases, identifying ≥ 2,000 variants underlying disease (1-3). Over 500 of these mutations are common and have been represented in multiple patients. An underrepresentation of specific ethnic groups in open access international DNA databases occurs, which highlights the importance of having DNA databases from ethnically matching healthy individuals. This is a feature that nation-specific human genome sequencing efforts such as the SHGP have achieved. However, nation-specific studies such as the SHGP are independent and do not form part of a global consortium or open access multipopulation DNA database, limiting the availability and global applications of the study findings. This differs from projects such as the 1000 Genomes Project that allows access to sequencing data from ≥ 26 countries to be globally available.

The initial aim of the SHGP was to read the genomes of affected individuals to identify the casual disease genes and their mutations. If tens of thousands of datasets are correctly analyzed, the major disease risk factors in the Saudi population can be identified (1-5). Entire genomes are sequenced when genes that cause human disease are to be identified in a Mendelian fashion (9). Sequencing of the coding regions using whole-exome sequencing (WES) has been successful and has clinical utility, particularly in the case of neurological disorders. However, WES typically uncovers a number of variants from which identifying causal variants can be challenging. Several tools have been developed to prioritize WES variants, but these require expert analysis. A leading concern of WES in clinical genomics is secondary findings, defined as variants that are potentially medical relevant but are missed as they are unrelated to the medical rationale for the test. This inability of WES to identify incidental/secondary findings is a commonly encountered problem (9, 16).

An alternative to WES is the use of gene panels, which are when a range of genes relevant to a particular phenotype are sequenced in patients presenting that phenotype. This approach provides a focused analysis of clinically relevant genes minimizing many WES challenges. Difficulties in this process arise from the design of an appropriate gene list for a given phenotype, as many diseases have variable clinical presentation. Indeed, gene panels developed across laboratories often differ for the same disease. This can be overcome by relaxing the clinical indications and the integration of supporting evidence to classify diseases into five categories as follows: (1) pathogenic, (2) likely to be pathogenic, (3) uncertain significance of variant, (4) likely benign, or (5) benign. These guidelines have improved disease classification; however, in practice, they remain challenging to standardize. In addition, a shortcoming of gene panels is their need to be updated with the discovery of new gene-disease associations. This is challenging as further disease mutants accumulate from large international datasets.

The Saudi Mendeliome Group defined 13 broad clinical themes in which ~3000 Mendelian genes were distributed (17). The power of this approach was highlighted by the success of the gene panel on the diagnosis of 2300 patients presenting with a range of medical and surgical diseases. Al-Mousa and coworkers also displayed the benefits of targeted NGS panels for primary immunodeficiency diseases, which were more sensitive and cost-effective than WES approaches, particularly in patients with an atypical presentation of known primary immunodeficiency genes (18). Mustafa and coworkers advanced these studies, demonstrating the suitability of a targeted AmpliSeq Inherited Disease Panel (IDP) (consisting of 328 genes underlying more than 700 inherited diseases) for the first-line screening of genetic diseases following clinical validation (19). It is therefore clear that the sequencing approach is dependent on the type of disorder and degree of genetic or phenotypic heterogeneity. WES has utility for gene discovery studies when exploring the genetic landscape of the population and in diseases with genetic or phenotypic heterogeneity. For those diseases in which little genetic information is available, gene panels have utility, particularly for clinical conditions with distinct and homogenous phenotypes, for which there is sufficient knowledge of their genetic basis.

Despite the obvious progress, knowledge gaps and areas of concern remain. A leading challenge remains that NGS technology is underutilized. Clinical genomics are primarily employed to detect SNVs and indels, but its application in the screening of CNVs, RNA or methylation status remains limited (16). NGS also fails to detect repeat expansion mutations or to quantify alleles. Experts have also expressed concerns over the small numbers of clinical geneticists and/or genetic counselors, as well as the lack of genomic knowledge among active clinicians. This results in a loss of confidence in the application of NGS data to clinical decision making.

Intellectual disability (ID) is a common developmental disorder characterized by congenital limitations in intellectual function. Advances in high-throughput whole-genome sequencing (WGS) and single-cell sequencing have increased the number of causative genes identified for this human disease. As the clinical features of ID display heterogeneity regarding their genetic causative factors, the characterization of ID has benefited from these advances. In SA, the most common cause of ID is genetic and due to the marriage of relatives (1). The SHGP has expanded the locus and allelic heterogeneity of ID in recent years, and the culmination of these studies demonstrates the power of positional mapping to reveal new and unusual mutational mechanisms (20-25). Anazi et al. (21) described the phenotypic and genetic findings of 105 patients originating from 68 families with ID, revealing several new disease variants. These genes included TRAK1, GTF3C3, SPTBN4 and NKX6-2 as well as novel variants in 14 other genes, including ANKHD1, ASTN2, ATP13A1, FMO4, MADD, MFSD11, NCKAP1, NFASC, PCDHGA10, PPP1R21, SLC12A2, SLK, STK32C and ZFAT (21). In particular, MADD and PCDHGA10 were compelling candidates as biallelic deleterious variants in two independent ID families were discovered (21). In other studies, Anazi et al. (20) prospectively used molecular karyotyping, multigene panels and whole-exome sequencing (WES) in a cohort of over 300 ID subjects and compared the data with the clinical evaluations of the patients. WES identified independent mutations in three new candidate ID genes, DENND5A, NEMF and DNHD1, and all patients harboring these mutations displayed comparable phenotypes. In addition, de novo and recessive variants in 32 other genes were identified, namely, MAMDC2, TUBAL3, CPNE6, KLHL24, USP2, PIP5K1A, UBE4A, TP53TG5, ATOH1, C16ORF90, SLC39A14, TRERF1, RGL1, CDH11, SYDE2, HIRA, FEZF2, PROCA1, PIANP, PLK2, QRFPR, AP3B2, NUDT2, UFC1, BTN3A2, TADA1, ARFGEF3, FAM160B1, ZMYM5, SLC45A1, ARHGAP33 and CAPS2 (20). Causal variants from previously published ID genes, including ASTN1, HELZ, THOC6, WDR45B, ADRA2B and CLIP1, were also confirmed, strengthening the accuracy of the study. The very recent identification of WDR45B as an ID gene in two large cohorts of affected individuals was revealed in independent studies (25). Additionally, KIF14 dysfunction was revealed as a cause of autosomal recessive primary microcephaly (an extremely rare condition characterized by a reduced cerebral cortex accompanied by ID) (26). KIF14 is a mitotic motor protein that is required for spindle localization of the mitotic citron rho-interacting kinase, CIT, highlighting a cellular mechanism of disease progression.

Genetic sequencing has also expanded our knowledge of ID formation. Patel and colleagues demonstrated variable expressivity in three consanguineous families linked to novel 3’ UTR mutations in SLC4A4, a gene known to be mutated in a syndromic form of ID (27). The 3’ UTR motif mediates posttranscriptional control of many genes, and a marked reduction in the transcript level of SLC4A4 was observed in cells from ID patients. This novel mutational mechanism expanded our knowledge of the variables that underlie phenotypic expressivity in human disease (27).

For many years, knowledge of the mutations in genes that lead to neurological disorders in SA has been lacking. The gaps in this knowledge have been filled using WES in multiple consanguineous families presenting diseases associated with impaired brain function. Alazami et al. (28) identified a total of 33 genes (SPDL1, TUBA3E, INO80, NID1, TSEN15, DMBX1, CLHC1, C12orf4, WDR93, ST7, MATN4, SEC24D, PCDHB4, PTPN23, TAF6, TBCK, FAM177A1, KIAA1109, MTSS1L, XIRP1, KCTD3, CHAF1B, ARV1, ISCA2, PTRH2, GEMIN4, MYOCD, PDPR, DPH1, NUP107, TMEM92, EPB41L4A, and FAM120AOS) as mutations that may impair brain activity, but in vitro and in vivo data that support a direct link between the identified mutations and impaired brain activity are required. Sowada et al. (29) described new forms of developmental and epileptic encephalopathies (DEE) with deleterious biallelic variants in PTPN23, a tyrosine kinase strongly expressed in neuronal tissue. The phenotype was characterized by early onset drug-resistant epilepsy, developmental delay, microcephaly, and on occasion premature death. Ramadan et al. (30) identified that recessive mutations in SCN1B can cause severe epilepsy in humans in addition to the known dominant SCN1B mutations, which were previously reported. This highlighted the need to consider recessive mutations in the interpretation of variants in typically dominant genes.

Shamseldin et al. (31) identified other candidate genes involved in a distinct neurodevelopmental phenotype in a multiplex consanguineous family. Mutations in MICU2, a major component of the mitochondrial calcium uniporter complex, lead to impaired mitochondrial Ca²⁺ homeostasis and severe cognitive impairment, spasticity, and white matter involvement (31). Al Mutairi and colleagues also identified mutations in the short-chain enoyl-CoA hydratase (SCEH), a mitochondrial enzyme involved in the oxidation of fatty acids and valine catabolism in early childhood Leigh syndrome (32). When these mutations were accompanied by a missense mutation in ECHS1, a lethal phenotype was observed.

The phenotypic and molecular spectrum of Aicardi-Goutières Syndrome (a rare genetic neurological disorder) was recently noted by Al-Mutairi and coworkers to be caused by homozygous mutations in RNASEH2B, RNASEH2A RNASEH2C, SAMHD1, and TREX1 as well as heterozygous mutations in IFIH1 in Arab pediatric patients (13). Additionally, in cases of spastic ataxia and hypomyelination, homozygosity mapping and WES revealed missense mutations in NKX6-2, which are known to encode a transcriptional repressor with high CNS expression (33). Accordingly, when NKX6-2 was silenced in mice, hypomyelination was observed (15, 33).

Studies assessing systemic lupus erythematosus (SLE) in Arab countries suggest its high prevalence in the Arab world (34-36). Using NGS, Carbonella et al. (37) identified a homozygous 2 bp deletion in DNASE1L3 that leads to autosomal recessive autoimmune disease (AID), which mimics SLE. The same mutations were reported in three siblings from consanguineous parents who presented with hypocomplementemic urticarial vasculitis syndrome (HUVS). As approximately 50% of individuals with HUVS develop SLE, whether SLE is a subphenotype or a separate condition was not clear. Studies to identify the cause of an autosomal-recessive form of systemic juvenile idiopathic arthritis (JIA), which displays both autoimmune- and auto-inflammatory-etiologies, in SA patients identified a homoallelic missense mutation in LACC1 (38). This gene encodes the enzyme laccase, which is a multicopper oxidoreductase. Missense mutations were identified using WES and confirmed by Sanger sequencing. The mutations were present in all investigated cases of JIA disease based on an autosomal-recessive pattern of inheritance. Given the known association of LACC1 with Crohn’s disease and leprosy, this further highlighted this enzyme as a drug target in genetic autoinflammatory disorders (38). This originates from data demonstrating that leprosy hijacks the immune system, manipulating defense mechanisms and causing them to attack neurons.

Inherited cystic kidney disorders commonly cause end-stage renal disease. To define the phenotype and genotype of cystic kidney disease in fetuses and neonates, Al-Hamed et al. (39) correlated antenatal- and postnatal-renal-ultrasound examination with targeted exon sequencing using a defined renal gene panel. The disease phenotypes observed were severe, with 36 cases of stillbirth or perinatal death. Renal gene panel testing identified causative mutations in up to 60% of the families, and mutations were found in 12 genes, including an inferred novel variant in NEK8 (39). Mutations in CC2D2A were the most common cause of antenatal cystic kidney disease and suspected ciliopathy. The renal gene panel from this study holds promise for a rapid molecular diagnosis for this disease in the future.

Meckel-Gruber syndrome (MKS) is characterized by occipital encephalocele, polydactyly and polycystic kidneys and is genetically heterogeneous with mutations in up to twelve genes known to date. Using a gene panel and NGS in 25 MKS families, Shaheen et al (40) identified a homozygous splice variant in TMEM107, which encodes for a protein that localizes to the transition zone at the proximal region of the ciliary axoneme. As TMEM107-deficient mice are known to display typical ciliopathy phenotypes, TMEM107 mutations are now accepted as a factor contributing to human MKS and thus represent a novel drug target for this disorder. In a large population of patients with phenotypes that span the entire ciliopathy spectrum, loss of function mutations in TXNDC15 encoding a thiol isomerase causing MKS were identified (41). Other novel candidate genes for ciliopathies were also discovered in these studies, including TRAPPC3, EXOC3L2, FAM98C, C17orf61, LRRCC1, NEK4, and CELSR2.

Hereditary cancer syndromes account for approximately 5% of all malignancies. Hereditary cancer syndromes occur due to an inherited mutation increasing the risk of tumor development, which typically occurs at an early age (42). In the majority of hereditary malignant cancers, the elevated risk is due to the mutation of a single gene (monogenic hereditary disease). The ability to identify germline variants in hereditary cancer cases has been a challenge. A history of incomplete cataloguing of cancer-relevant genes combined with a lack of agreement on the patients who should be tested has posed significant barriers. In a recent study from SA, Siraj et al. (43) designed a hereditary oncogenesis predisposition evaluation (HOPE) that consisted of many of the genes with known associations with cancer and assessed the effectiveness of HOPE on ≥ 1300 tumor/blood samples from patients with ovarian, breast, colorectal and thyroid cancer. Most notably, pathogenic alleles in DNA repair/genomic instability genes other than BRCA2, ATM and PALB2 accounted for at least 11.1, 16.8, 50 and 45.5% of mutation-positive ovarian, breast, thyroid and colorectal cancer patients (CRC), respectively (43). A family history was absent in many of these mutation-positive cases, suggesting a high contribution of germline mutations to cancer predisposition and extending our knowledge beyond “classic” hereditary cancer genes.

Embryonic lethality is recognized as a phenotypic result of individual gene mutations. However, identifying embryonically lethal genes in humans is challenging, particularly when the phenotype manifests at the preimplantation stage. To catalogue recessively acting embryonic lethal genes in SA, Alazami et al. (44) identified two families with a female-limited infertility phenotype. Using autozygosity mapping and WES, the phenotype was matched to a single mutation in TLE6, a maternal-effect gene that encodes a member of the subcortical maternal complex (44). Female patients homozygous for TLE6 mutations were found to not undergo early cleavage, leading to sterility. The human mutation abrogates TLE6 phosphorylation, which is reported to be critical for oocyte meiosis (44). This was the first report to identify a human defect in the subcortical maternal complex. Maddirevula et al. (45) performed positional mapping and WES and revealed two homozygous deleterious variants in PATL2, which is also necessary for female meiosis and fertility. PATL2 encodes a highly conserved oocyte-specific mRNP repressor of translation, again identifying a novel disease mechanism in humans. The diagnosis of these mutations in IVF clinics may thus hold great promise for the identification of fetal abnormalities.

Orofacial clefting is among the most prevalent type of birth defect. Mutations in the HYAL2 gene encoding hyaluronidase 2 (which degrades extracellular hyaluronan) were identified as a critical component of the developing heart and palatal shelf matrix (46). In SA studies, Harms et al. (47) identified a homozygous truncating variant in CDH11 associated with Elsahy–Waters syndrome (EWS), also known as branchial–skeletal–genital syndrome. Included in the clinical features of CDH11, mutation-positive individuals had upper eyelid coloboma, which is a new phenotype of this disease that may aid in its diagnosis.

In other studies, Shaheen et al. (48) identified loss-of-function mutations in SMG9 that lead to a loss of nonsense-mediated decay (NMD), which is an important process to degrade transcripts containing premature stop codons that removes their potentially harmful consequences. SMG9 was shown to be required for normal human development, most likely through this transcriptional regulatory role.

Congenital hydrocephalus is an important birth defect that results in children born with an excessive accumulation of cerebrospinal fluid in the brain. WES combined with positional mapping identified causal mutations in 16 genes; interestingly, none of these mutations were X-linked. Ciliopathies and dystroglycanopathies were the most common etiologies of this congenital hydrocephalus cohort. In a single family with four affected members, a homozygous truncating variant in EML1 was identified (49). Recessive mutations in WDR81, previously linked to cerebellar ataxia, mental retardation, and disequilibrium syndrome, were also found. Other previously identified candidates, including MPDZ, were identified, thus highlighting the importance of recessive mutations in this disease condition (49).

Congenital disorders of glycosylation are frequently associated with muscle weakness in the Arab population. Using WES, Monies and coworkers reported clinical and pathological features resulting from a homozygous mutation of ALG2 in an extended family (50). Mutations of ALG2 manifested as limb and muscle wasting, with defects at both the neuromuscular junction and sarcomere evident. The same group developed a first-line diagnostic assay for limb-girdle muscular dystrophy and myopathies by evaluating a panel of 759 OMIM genes associated with neurological disorders in Saudi patients presenting with muscle weakness (51).

Larsen syndrome (LS) is characterized by the dislocation of large joints with heterozygous FLNB mutations, accounting for the majority of cases. However, biallelic mutations in CHST3 and B4GALT7 were recently identified by Patel et al. (52), confirming recessive forms of the disease. In a multiplex consanguineous Saudi family affected by severe and recurrent large joint dislocation and severe myopia, a homozygous truncating variant was identified in GZF1 through a combined autozygome and exome approach (52). The same approach identified a second homozygous truncating GZF1 variant in a second multiplex consanguineous family affected by severe myopia and milder skeletal involvement (53). GZF1 encodes the GDNF-inducible zinc finger protein 1, which is a transcription factor of unknown developmental function, and it is expressed in the eyes and limbs of developing mice. These studies reveal new functionality of this protein and its contribution to human disease when dysfunctional.

The spectrum of genetic muscle diseases also extends to diseases of the heart muscle. Dilated cardiomyopathy (DCM) is a common cardiomyopathy that leads to systolic dysfunction and heart failure due to the heart becoming enlarged, thick and rigid. Despite knowledge of over 30 known genes for DCM (primarily in sarcomere and cytoskeletal proteins), Al-Yacoub et al. (54) identified mutations in FBXO32 as a novel DCM-causing locus. The protein encoded by FBXO32 (also known as atrogin-1) is an E3 ligase that is expressed selectively in skeletal muscle and cardiomyocytes.

Molecular genetic studies are of increasing importance in the diagnosis and classification of several other diseases. Causative mutations in SKIV2L and TTC37 that cause congenital diarrheal disorders in SA patients were identified by Monies and coworkers (55). Congenital cafe-au-lait spots on the pelvis and lower limbs were a unique and consistent clinical feature of these patients and will aid the differential diagnosis of congenital diarrheal disorders in the future (55).

Ectodermal dysplasia is a highly heterogeneous group of disorders that affect skin, hair, nails and teeth. In an autosomal dominant form of ectodermal dysplasia, Shamseldin et al. (53) identified a novel variant in KDF1 in a Saudi family. The recapitulation of the phenotype was observed in KDF1 knockout mice, suggesting a causal role played by the KDF1 variant (56).

Diabetic retinopathy (DR) is a common clinical expression of diabetes mellitus-induced vasculopathy and is a major cause of vision loss. In a cohort of SA diabetic individuals who did not develop DR 10 years after diagnosis compared to SA diabetic individuals with DR, WES identified three genes (NME3, LOC728699, and FASTK) whose rare variant burden protects against DR, highlighting them as attractive candidate drug targets (57). In syndromic RD cases, WES revealed AGBL5 and CDH16 as likely gene candidates for the disease (58). A homozygous truncating mutation in DNAJC17 in a family with retinitis pigmentosa and hypogammaglobulinemia was also identified, thus expanding the allelic spectrum of known RD genes. Patel et al. (56), when investigating the genetic basis of pediatric cataracts, applied a multigene panel as well as WES and identified 15 novel genes, including GEMIN4, which was mutated in families with cataracts and global developmental delay. Mutations in RIC1 were identified in patients with cataracts, brain atrophy and microcephaly. Among others, two further candidates that were biallelically inactivated in single cataract families were reported, including TAF1A (cataract with global developmental delay) and WDR87 (nonsyndromic cataract) (56). This expanded the allelic and locus heterogeneity of pediatric cataracts for future diagnosis/treatment.

Advanced cholestatic liver disease is a leading referral to pediatric liver transplant centers in Arab countries. Using an NGS panel of pediatric patients with advanced cholestatic liver disease, causal mutations were identified in the majority of cases and included the novel identification of TJP2 and VIPAS39 (59). Due to the alarmingly high carrier frequency of founder mutations identified in this cohort, primary prevention through carrier screening was suggested as the most effective method of minimizing this disease incidence.