A systematic review on machine learning approaches in the diagnosis and prognosis of rare genetic diseases

BACKGROUND

The diagnosis of rare genetic diseases is often challenging due to the complexity of the genetic underpinnings of these conditions and the limited availability of diagnostic tools. Machine learning (ML) algorithms have the potential to improve the accuracy and speed of diagnosis by analyzing large amounts of genomic data and identifying complex multiallelic patterns that may be associated with specific diseases. In this systematic review, we aimed to identify the methodological trends and the ML application areas in rare genetic diseases.

METHODS

We performed a systematic review of the literature following the PRISMA guidelines to search studies that used ML approaches to enhance the diagnosis of rare genetic diseases. Studies that used DNA-based sequencing data and a variety of ML algorithms were included, summarized, and analyzed using bibliometric methods, visualization tools, and a feature co-occurrence analysis.

FINDINGS

Our search identified 22 studies that met the inclusion criteria. We found that exome sequencing was the most frequently used sequencing technology (59%), and rare neoplastic diseases were the most prevalent disease scenario (59%). In rare neoplasms, the most frequent applications of ML models were the differential diagnosis or stratification of patients (38.5%) and the identification of somatic mutations (30.8%). In other rare diseases, the most frequent goals were the prioritization of rare variants or genes (55.5%) and the identification of biallelic or digenic inheritance (33.3%). The most employed method was the random forest algorithm (54.5%). In addition, the features of the datasets needed for training these algorithms were distinctive depending on the goal pursued, including the mutational load in each gene for the differential diagnosis of patients, or the combination of genotype features and sequence-derived features (such as GC-content) for the identification of somatic mutations.

CONCLUSIONS

ML algorithms based on sequencing data are mainly used for the diagnosis of rare neoplastic diseases, with random forest being the most common approach. We identified key features in the datasets used for training these ML models according to the objective pursued. These features can support the development of future ML models in the diagnosis of rare genetic diseases.

Rare Coding Variants in Patients with Non-Syndromic Vestibular Dysfunction

Vertigo due to vestibular dysfunction is rare in children. The elucidation of its etiology will improve clinical management and the quality of life of patients. Genes for vestibular dysfunction were previously identified in patients with both hearing loss and vertigo. This study aimed to identify rare, coding variants in children with peripheral vertigo but no hearing loss, and in patients with potentially overlapping phenotypes, namely, Meniere's disease or idiopathic scoliosis. Rare variants were selected from the exome sequence data of 5 American children with vertigo, 226 Spanish patients with Meniere's disease, and 38 European-American probands with scoliosis. In children with vertigo, 17 variants were found in 15 genes involved in migraine, musculoskeletal phenotypes, and vestibular development. Three genes, OTOP1, HMX3, and LAMA2, have knockout mouse models for vestibular dysfunction. Moreover, HMX3 and LAMA2 were expressed in human vestibular tissues. Rare variants within ECM1, OTOP1, and OTOP2 were each identified in three adult patients with Meniere's disease. Additionally, an OTOP1 variant was identified in 11 adolescents with lateral semicircular canal asymmetry, 10 of whom have scoliosis. We hypothesize that peripheral vestibular dysfunction in children may be due to multiple rare variants within genes that are involved in the inner ear structure, migraine, and musculoskeletal disease.

Using coding and non-coding rare variants to target candidate genes in patients with severe tinnitus

Tinnitus is the phantom percept of an internal non-verbal set of noises and tones. It is reported by 15% of the population and it is usually associated with hearing and/or brain disorders. The role of structural variants (SVs) in coding and non-coding regions has not been investigated in patients with severe tinnitus. In this study, we performed whole-genome sequencing in 97 unrelated Swedish individuals with chronic tinnitus (TIGER cohort). Rare single nucleotide variants (SNV), large structural variants (LSV), and copy number variations (CNV) were retrieved to perform a gene enrichment analysis in TIGER and in a subgroup of patients with severe tinnitus (SEVTIN, n = 34), according to the tinnitus handicap inventory (THI) scores. An independent exome sequencing dataset of 147 Swedish tinnitus patients was used as a replication cohort (JAGUAR cohort) and population-specific datasets from Sweden (SweGen) and Non-Finish Europeans (NFE) from gnomAD were used as control groups. SEVTIN patients showed a higher prevalence of hyperacusis, hearing loss, and anxiety when they were compared to individuals in the TIGER cohort. We found an enrichment of rare missense variants in 6 and 8 high-constraint genes in SEVTIN and TIGER cohorts, respectively. Of note, an enrichment of missense variants was found in the CACNA1E gene in both SEVTIN and TIGER. We replicated the burden of missense variants in 9 high-constrained genes in the JAGUAR cohort, including the gene NAV2, when data were compared with NFE. Moreover, LSVs in constrained regions overlapping CACNA1E, NAV2, and TMEM132D genes were observed in TIGER and SEVTIN.

Rare coding variants involving MYO7A and other genes encoding stereocilia link proteins in familial meniere disease

The MYO7A gene encodes a motor protein with a key role in the organization of stereocilia in auditory and vestibular hair cells. Rare variants in the MYO7A (myosin VIIA) gene may cause autosomal dominant (AD) or autosomal recessive (AR) sensorineural hearing loss (SNHL) accompanied by vestibular dysfunction or retinitis pigmentosa (Usher syndrome type 1B). Familial Meniere's disease (MD) is a rare inner ear syndrome mainly characterized by low-frequency sensorineural hearing loss and episodic vertigo associated with tinnitus. Familial aggregation has been found in 6-8% of sporadic cases, and most of the reported genes were involved in single families. Thus, this study aimed to search for relevant genes not previously linked to familial MD. Through exome sequencing and segregation analysis in 62 MD families, we have found a total of 1 novel and 8 rare heterozygous variants in the MYO7A gene in 9 non-related families. Carriers of rare variants in MYO7A showed autosomal dominant or autosomal recessive SNHL in familial MD. Additionally, some novel and rare variants in other genes involved in the organization of the stereocilia links such as CDH23, PCDH15 or ADGRV1 co-segregated in the same patients. Our findings reveal a co-segregation of rare variants in the MYO7A gene and other structural myosin VIIA binding proteins involved in the tip and ankle links of the hair cell stereocilia. We suggest that recessive digenic inheritance involving these genes could affect the ultrastructure of the stereocilia links in familial MD.

Burden of Rare Variants in the OTOG Gene in Familial Meniere's Disease

OBJECTIVES

Meniere's disease (MD) is a rare inner ear disorder characterized by sensorineural hearing loss, episodic vertigo, and tinnitus. Familial MD has been reported in 6 to 9% of sporadic cases, and few genes including FAM136A, DTNA, PRKCB, SEMA3D, and DPT have been involved in single families, suggesting genetic heterogeneity. In this study, the authors recruited 46 families with MD to search for relevant candidate genes for hearing loss in familial MD.

DESIGN

Exome sequencing data from MD patients were analyzed to search for rare variants in hearing loss genes in a case-control study. A total of 109 patients with MD (73 familial cases and 36 early-onset sporadic patients) diagnosed according to the diagnostic criteria defined by the Barany Society were recruited in 11 hospitals. The allelic frequencies of rare variants in hearing loss genes were calculated in individuals with familial MD. A single rare variant analysis and a gene burden analysis (GBA) were conducted in the dataset selecting 1 patient from each family. Allelic frequencies from European and Spanish reference datasets were used as controls.

RESULTS

A total of 5136 single-nucleotide variants in hearing loss genes were considered for single rare variant analysis in familial MD cases, but only 1 heterozygous likely pathogenic variant in the OTOG gene (rs552304627) was found in 2 unrelated families. The gene burden analysis found an enrichment of rare missense variants in the OTOG gene in familial MD. So, 15 of 46 families (33%) showed at least 1 rare missense variant in the OTOG gene, suggesting a key role in familial MD.

CONCLUSIONS

The authors found an enrichment of multiplex rare missense variants in the OTOG gene in familial MD. This finding supports OTOG as a relevant gene in familial MD and set the groundwork for genetic testing in MD.

Systematic Review of Sequencing Studies and Gene Expression Profiling in Familial Meniere Disease

Familial Meniere Disease (FMD) is a rare inner ear disorder characterized by episodic vertigo associated with sensorineural hearing loss, tinnitus and/or aural fullness. We conducted a systematic review to find sequencing studies segregating rare variants in FMD to obtain evidence to support candidate genes for MD. After evaluating the quality of the retrieved records, eight studies were selected to carry out a quantitative synthesis. These articles described 20 single nucleotide variants (SNVs) in 11 genes (FAM136A, DTNA, PRKCB, COCH, DPT, SEMA3D, STRC, HMX2, TMEM55B, OTOG and LSAMP), most of them in singular families-the exception being the OTOG gene. Furthermore, we analyzed the pathogenicity of each SNV and compared its allelic frequency with reference datasets to evaluate its role in the pathogenesis of FMD. By retrieving gene expression data in these genes from different databases, we could classify them according to their gene expression in neural or inner ear tissues. Finally, we evaluated the pattern of inheritance to conclude which genes show an autosomal dominant (AD) or autosomal recessive (AR) inheritance in FMD.

Clinical and molecular genetics of Meniere disease

Meniere disease (MD) represents a heterogeneous group of relatively rare disorders of the inner ear that causes vertigo attacks, fluctuating sensorineural hearing loss (SNHL) involving low and medium frequencies, tinnitus, and aural fullness. MD has been attributed to an accumulation of endolymph in the cochlear duct. The diagnosis of MD is based on the phenomenological association of clinical symptoms including SNHL during the vertigo attacks. At least two mechanisms are involved in MD: (a) a pro-inflammatory immune response mediated by interleukin-1 beta (IL-1β), tumor necrosis factor alpha (TNFα), and IL-6, and (b) nuclear factor-kappa B (NF-κB)-mediated inflammation in the carriers of the single nucleotide variant rs4947296. The majority of MD cases are considered sporadic, although familial aggregation has been recognized in European and East Asian populations in multiplex families, supporting a genetic contribution to the disease. In sporadic MD cases, the main genetic findings involve multiplex rare variants in several SNHL genes, such as GJB2, USH1G, SLC26A4, ESRRB, and CLDN14, and axonal guidance signaling genes, such as NTN4 and NOX3. Familial aggregation has been reported in 6–8 % of MD cases, and most families show an autosomal dominant inheritance. Few rare missense heterozygous variants have been described in simplex families in six genes (COCH, FAM136A, DTNA, PRKCB, SEMA3D, and DPT). Of note, 33 % of familial MD individuals show singleton and multiplex rare missense variants in the OTOG gene, suggesting a multiallelic inheritance. Moreover, potentially pathogenic rare variants in the familial genes FAM136A, DTNA, and DPT have been reported in Korean singletons with sporadic MD. Rare variants may have a significant contribution to sporadic and familial MD. The interaction of common cis-regulatory variants located in non-coding regions and rare variants in coding regions in one or more genes will determine the variation on the phenotype in MD. Further studies on genotype–phenotype correlations are required to improve the yield of genetic diagnosis, and different types of variants seem to contribute to the genetic structure of MD.

Enrichment of damaging missense variants in genes related with axonal guidance signalling in sporadic Meniere's disease

INTRODUCTION

Meniere's disease (MD) is a rare inner ear disorder with a significant genetic contribution defined by a core phenotype: episodic vertigo, sensorineural hearing loss and tinnitus. It has been mostly described in sporadic cases, familial cases being around 10% of the observed individuals. It is associated with an accumulation of endolymph in the inner ear, but the molecular underpinnings remain largely unknown. The main molecular pathways showing higher differentially expressed genes in the supporting cells of the inner ear are related to cochlea-vestibular innervation, cell adhesion and leucocyte extravasation. In this study, our objective is to find a burden of rare variants in genes that interact with the main signalling pathways in supporting cells of the inner ear in patients with sporadic MD.

METHODS

We designed a targeted-sequencing panel including genes related with the main molecular pathways in supporting cells and sequenced 860 Spanish patients with sporadic MD. Variants with minor allele frequencies <0.1 in the gene panel were compared with three independent reference datasets. Variants were classified as loss of function, missense and synonymous. Missense variants with a combined annotation-dependent depletion score of >20 were classified as damaging missense variants.

RESULTS

We have observed a significant burden of damaging missense variants in few key genes, including the NTN4 gene, associated with axon guidance signalling pathways in patients with sporadic MD. We have also identified active subnetworks having an enrichment of rare variants in sporadic MD.

CONCLUSION

The burden of missense variants in the NTN4 gene suggests that axonal guidance signalling could be a novel pathway involved in sporadic MD.

Excess of Rare Missense Variants in Hearing Loss Genes in Sporadic Meniere Disease

Meniere's disease (MD) is a clinical spectrum of rare disorders characterized by vertigo attacks, associated with sensorineural hearing loss (SNHL) and tinnitus involving low to medium frequencies. Although it shows familial aggregation with incomplete phenotypic forms and variable expressivity, most cases are considered sporadic. The aim of this study was to investigate the burden for rare variation in SNHL genes in patients with sporadic MD. We conducted a targeted-sequencing study including SNHL and familial MD genes in 890 MD patients to compare the frequency of rare variants in cases using three independent public datasets as controls. Patients with sporadic MD showed a significant enrichment of missense variants in SNHL genes that was not found in the controls. The list of genes includes GJB2, USH1G, SLC26A4, ESRRB, and CLDN14. A rare synonymous variant with unknown significance was found in the MARVELD2 gene in several unrelated patients with MD. There is a burden of rare variation in certain SNHL genes in sporadic MD. Furthermore, the interaction of common and rare variants in SNHL genes may have an additive effect on MD phenotype. This study will contribute to design a gene panel for the genetic diagnosis of MD.