Central resources of variant discovery and annotation and its role in precision medicine

Locus Reference Genomic (LRG) (URL: http://www.lrg-sequence.org) was established to curate the variants with clinical implications and contains stable genomic DNA sequences, transcript, and protein reference sequences for reporting clinically relevant sequence variants [8]. The updatable section contains the mapping of the LRG to the most recent genome build and the most advanced and up-to-date biological knowledge for the LRG region from Ensembl and NCBI. To ensure that each LRG record is updated, the curators review all existing LRGs twice a year and when a new genome build is released.

The LRG is aligned to the reference genome to determine its mapping coordinates. Each area of contiguous alignment is described separately, along with any differences between the LRG and the reference. The existence of LRG is important for researchers to understand the function of a gene and how variations might affect its function and cause clinical implications [2]. The reference genome is not from a single person only. It is derived from the DNA of certain people, as data sequenced from thousands of genomes will show a drastic variation in DNA content. Among the advantages of having a reference is that the genome would be useful in analyses employing DNA testing for forensic purposes, as well as paternity determination, through the repetitive sequence of variant regions.

An individual's DNA varying from those of other individuals is caused by differences in ancestral pattern, unique traits, and susceptibility to certain diseases [9]. Therefore, a greater understanding of variant annotation and interpretation is essential and contributes significant roles to further determining how this part of the genome influences life and how knowledge of these variants can be applied toward optimal health management and care [3].

Variant annotation is an important step for the precise determination of sequence variants, as genetic alterations may result in various effects on genes and proteins, including a gain of function mutation that commonly affects the protein by introducing new protein functions such as missense mutation. Additionally, genetic alteration may also be caused by loss of function mutation, which could abolish the protein function. The effect may result in nonsense mutation that terminates the proteins or splice site mutation that can disrupt RNA splicing, causing alteration to protein sequence. By annotating the variants according to their functional classification, it would be easier for researchers to prioritize their target variants and perform further interpretation.

Curation of the human genome is a time-consuming process because it depends on unstructured sources such as published literature, collections of laboratory clinical data, disease-specific groups, and databases of functional experiments. While the genome sequencing revolution has led to the sequencing and assembly of many thousands of new genomes, genome annotation still uses very nearly the same technology that we have used for the past 2 decades. Variant identification and annotation are pivotal starting points for an accurate variant interpretation. The prevailing challenge in the variant annotation is the conversion of genomic variant coordinates to their corresponding localization of cDNA or amino acid [10]. Genome interpretation is highly dependent on the well-defined genomic variant coordinates, as the difference in the alignment of sequence variants may lead to incorrect variant nomenclature. An accurate variant interpretation is also determined by a consistent variant representation using a correct transcript accession [10].

Additionally, issues and arguments of data inconsistency may arise due to disparities between the laboratories and consequently cause difficulties for clinicians in fully utilizing genetic information for their patients’ management. To successfully resolve these discrepancies, broad, up-to-date, and flexible standards that are best equipped and in line with the expanding genomics information should be deployed, so as to ensure the benefit of overcoming any argument in clinical classifications [11, 12]. Additionally, various variant detection software tools can strictly identify a specific alteration, including SNV, indels, structural variants, and CNV. Clinical laboratories must also understand and be aware of any limitations of the variant detection tools. The quality of the results from each laboratory should be maintained through an appropriate laboratory validation of the bioinformatics pipeline and the use of commercially purchased bioinformatics tools and software [11, 12].

Another critical point to be considered is the inclusion of significant clinical information and interpretation of the identified variants such as the gene symbol, the location of the variant, type of variant, correct HGVS nomenclature, and predicted protein sequence alterations. An additional source of information, such as cross-references to cancer-specific databases, multiple genomic databases, population frequency databases, and pre-computed in silico-based prediction tools, is also very useful in assisting in variant interpretation [10].

Variant annotation is greatly dependent on computational predictions to characterize the effect of the variant on a molecular level. While the use of this prediction tool was aided by the availability of published reports and an expert panel for validation, few limitations still exist in these tools. For annotation tools such as SIFT and PolyPhen that predict the functional effect of SNP on protein and phenotype, these tools have their own limitations. SIFT could not predict the protein structure while PolyPhen includes that information; however, the accuracy of this tool depends on the structural class of the protein [3, 13]. Various databases have now provided gene expression data in addition to genome annotation and allele frequencies for population-based information [14, 15]. However, these databases are still characterized by a few limitations, including an absence of data integration for different kinds of information, and an example of this particular limitation would be that, for a given DNA variant query at any one time, gene expression data are not available in a state of incorporation within the functional prediction and pathogenicity classification [16]. Most of these databases also only allow a limited number of gene and DNA variant queries [17]. Massive data for variant interpretation has been a major challenge for researchers and the integration of different information from the available resources to support the evaluation of evidence for data interpretation will be a better solution to resolve this limitation. At this point, variant annotation from multiple annotation tools and databases must be considered for a complete variant interpretation [18].

Selected resources, as described in Table 1, show the valuable information and analysis tools needed to portray genetic variations along the proof continuum [19]. These public resources consist of variation data from both normal and affected populations [20]. Advancements in innovation and investigation abilities, resulting in costs’ and time consumption reduction, have alternately changed genomics, empowering one to pinpoint applicable genetic variants inside a large number of variations in a sequenced database [21]. However, recovering factual and useful annotation important at the single nucleotide level is still problematic [22].

ACMG Guidelines

In 2015, the American College of Medical Genetics and Genomics (ACMG) and the Association for Molecular Pathology (AMP) released joint recommendations for interpreting variants in genes associated with Mendelian disorders. These recommendations are part of an effort to establish a common framework for variant classification based on a standardized and transparent assessment of different lines of evidence. ACMG Guidelines provide several levels of evidence to ascertain the clinical significance of each genomic variant. However, there is still a challenging task to determine the pathogenicity of variants, especially in cancer. Several factors and information need to be taken into account as each genomic alteration can have an extent of much clinical efficacy, such as the diagnosis, prognosis, therapy selection, and monitoring, though only a few cancers have strong evidence of pathogenicity [69]. Until now, peer-reviewed literature, guidelines for clinical practice, and any relevant mutation database are being utilized as the main sources in equipping these variants with useful clinically significant evidence for further assessment [70].

The ACMG–AMP Guidelines propose the use of specific terminology, namely “pathogenic,” “likely pathogenic,” “uncertain significance,” “likely benign,” and “benign,” to characterize variants in Mendelian disorders (Figure 1). The variants are classified into these 5 categories according to different types of evidence including population data, computational data, functional data, and segregation data. This framework defined 28 evidence criteria organized by type and strength, which are used to classify the variants into the 5 categories. The framework proposed by the guidelines, encompassing various genes, diseases, and inheritance patterns, is intended for general use; however, to effectively execute the process of variant interpretation, expert judgment in evaluating and weighing evidence is needed.” By using the guidelines, a mutation is characterized by a change in the DNA base sequence, whereas a polymorphism is characterized by a frequency of greater than 1%. Both mutation and polymorphism were later changed to “variant” to avoid confusion arising from inaccurate assumptions of pathogenic or benign effects [58].

The Clinical Genome Resource (ClinGen) convenes all Variant Curation Expert Panels (VCEPs) to perform gene- and disease-specific modifications to the ACMG–AMP framework for variants related to a particular disease and inheritance pattern [32]. Additionally, members of ClinGen have formed a working group, called the Sequence Variant Interpretation Working Group (SVI WG). This working group aims to provide refinements of ACMG–AMP guidelines across multiple domains and to harmonize guidelines made by individual VCEPs [71]. SVI WG has published these recommendations (https://www.clinicalgenome.org/svi/).

Figure 1.

Figure 2.

The current effort of ClinGen is the development of the ClinGen Variant Curation Interface (VCI). The ClinGen VCI is a global open-source variant classification platform that helps with the use of evidence criteria and the classification of variants based on the ACMG–AMP variant classification guidelines. There are 43 VCEPs and the majority are in the process of the establishment (https://clinicalgenome.org/affiliation/vcep/#ep_table_heading). Figure 2 indicates the parameters used by the ClinGen Hemoglobinopathy VCEP for annotation variants in disorders of hemoglobinopathies.

Although these modifiers may not fully represent human phenotypes, they comprise a 5-tier system of classification for variants that are significant to Mendelian disease, which should be supported with a clinical condition and inheritance pattern.

Precision Medicine Initiative (PMI) was initiated in 2015 to accelerate and enhance the understanding of precision medicine and to take into consideration factors such as genetics, physiology, and environmental parameters for ascertaining the best approach to disease treatment for any individual patient [72]. Previous and current medical practices usually do not consider this biomedical information due to inaccessible and insufficient sources. Additionally, genome sequencing and extensive biomedical data collection are usually still not an option due to the costs involved and hence not available for all [73, 74]. However, recently, high-throughput DNA sequencing technologies are starting to have an impact on various medical fields and it has become feasible to gather a substantial amount of health information data from patients to be applied for individualized medical treatment [75].

Precision medicine will aid physicians to choose the best treatment and prevention approach for each patient according to their personal health information and their genetic makeup [76]. Certain diseases are associated with and influenced by genetic variations as well as patients’ responses to certain medications. Therefore, the most suitable treatment can be considered to improve the patients’ outcomes. For instance, the finding of hundreds of genetic susceptibility loci in inflammatory bowel disease (IBD) patients contributes to better management of IBD [77]. The development of precision medicine improves our healthcare surroundings by providing precise knowledge of current trends of morbidity and mortality in populations or groups with specific diseases [74].

Various efforts toward precision medicine have been facilitated by many national and international public databases that classify and annotate genomic variation [78], including database for single nucleotide polymorphism (dbSNP) [29], ClinVar [7], and Online Mendelian Inheritance in Man (OMIM) [79]. The basis of precision medicine lies in cataloging genomic variation data of local genomes. Strong local reference databases can make precision medicine meaningful and the inadequacy of local reference data is the biggest challenge in promoting precision medicine in Asia.

Additionally, a decision from expert panels for variant annotation might be one of the challenges in precision medicine. Experts disclose that the disadvantages of personalized medicine are classifiable into 3 recognized issues, which are validity uncertainty, equity issues, and implementation [80]. Therefore, the availability of artificial intelligence (AI) is critical for the implementation of precision medicine. When AI is introduced into precision medicine, organizations can implement precision medicine using their own innovative approach. For instance, AI controls deep learning approaches to defeat the difficulties involved in massive data sets and unstructured data. To illustrate, Gubatan et al. [81] explained the most recent advanced technologies of AI in disease diagnosis, risk prediction, and disease severity evaluation, as well as prediction of clinical decision-making in IBD patients. Again, Jiang et al. [82] developed predictive models learned from historical data for predicting COVID-19 severity. Therefore, we deduce that variant annotation can be implemented into AI to assist in precision medicine in the future.

In clinical settings, AI facilitates clinicians to work more efficiently to make diagnoses more accurate. Additionally, AI helps companies in drug development to implement time and cost-effective measures as well as prevent errors in the system.

Globally, there are many initiatives to predict, with a high level of accuracy, which treatment and prevention strategies would be an appropriate choice for any particular disease that may present in a population. A primary challenge for clinical laboratories is to determine, with adequate evidence, the genes that can be associated with clinical findings and patients’ phenotypes. Therefore, ClinGen developed a method to help the community in this effort to assess the strength of evidence for each gene in specific diseases. ClinGen (clinicalgenome.org) is a multi-institutional initiative to standardize gene and variant curation as well as to improve the utility of genomics in medicine. ClinGen developed a method to evaluate the strength of evidence for providing supporting claims for every gene that causes particular diseases. Additionally, this effort was purposely initiated to establish an authoritative central resource for the characterization of clinically relevant genes and variants for precision medicine and future research [32].

ClinGen gene and variant curation is currently organized into 8 Clinical Domain Working Groups (CDWG) to emphasize on curation and standardization efforts in clarification. These CDWG include Cardiovascular, Hearing Loss, Hemostasis or Thrombosis, Hereditary Cancer, Inborn Errors of Metabolism, Neurodevelopmental, Neuromuscular, and RASopathies (mutations in genes of the Ras-MAPK pathway). Each CDWG manages numerous Gene or Variant Curation Expert Panels (GCEPs or VCEPs) performing curation activities on gene sets and diseases within the domain group [83].

As the ACMG–AMP Variant Interpretation Guidelines clarify and specify, ClinGen VCI curates the variants and expert review, and the VCEPs can provide interpreted variant sets to ClinVar and the ClinGen evidence repository (eRepo; erepo.clinicalgenome.org). This step encourages the sharing of data from diagnostic laboratories, creation of consensus interpretations, and development of a collection of gene-specified ACMG–AMP evidence codes, thereby paving the way for eventually ensuring more accurate interpretations for clinical use [75].

However, the critically important factor would also depend on ancestry and population reference data that are supported by previous data, developed through a detailed clinical investigation of affected patients and at-risk family members [84]. As the composition of pathogenic variants differs greatly among human populations, most data focused mainly on populations of European ancestry. In contrast, little is known about pathogenic DNA variants among populations from other parts of the world. This also includes those non-annotated variants that could be pathogenic in the minority of populations [85].

In the era of precision medicine, variant discovery and annotation are required for diagnosis, prevention, and therapy improvement, based on an individually tailored approach; however, progress is still slow and the application of precision medicine in patients’ treatment still needs enhancement. Despite the abundance of raw data, much genetic variation in the human genome still remains uncharacterized. For precision medicine initiatives to progress, it is crucial that the discovery and annotation pipeline for novel variants also be incorporated during the analysis and interrogation of genomic data. Providing a uniform set of annotation resources for certain genetic diseases will ease comparisons and the interpretation of results. It is hoped that the discovery of novel genomic variants and their annotations would accelerate the development of precision medicine, thus benefitting mankind.