Functional genomics uses genomic and other ‘omic information to better understand the function and regulation of genes, and how they operate in a dynamic system. Functional genomics can therefore give valuable insights into the implications of genetic variation for health and disease. In two new policy briefings we outline what functional genomics is, the state of the science and the potential applications of functional genomic approaches for genomic medicine. Here, we explore in more detail one promising example of functional genomics applications – clinical diagnosis of rare diseases.
Whole genome and whole exome sequencing (WGS and WES, respectively) are increasingly employed for the diagnosis of rare diseases with a suspected genetic cause. Despite the success of these approaches in uncovering causative genetic variants, around two thirds of patients with a rare genetic disease remain without a confirmatory diagnosis. This is partly due to challenges and limitations in understanding how genomic variation impacts gene regulation and expression, and the proteins. Studies are investigating the use of multiple ‘omics datasets to bolster information that can be gained by analysing the genetic material alone to understand functional aspects of disease.
What is functional genomics?
Functional genomics is a field of molecular biology that employs multiple ‘omics datasets from different “layers” of biological activity to understand the complex relationship between our genes and the traits we express. Different datasets can be used to understand changes in gene expression (transcriptomics), regulation of gene expression (epigenomics), the abundance of proteins, their structure and interactions (proteomics), and the quantities and diversity of metabolites (metabolomics).
Analysing multiple layers of biological activity together can give a more holistic picture of what is occurring on a molecular level and in particular for disease, what is going wrong.
New technology, new hope
Patients with rare disease face unique challenges when it comes to obtaining a diagnosis. Sequences generated by WES and WGS are difficult to interpret due to the sheer amount of variation between each person’s genome. Determining which variants contribute to the cause of disease requires a series of computational prioritisation steps to narrow down the thousands of variants into those that are more likely to cause disease (pathogenic). The variants that are left can then be manually or computationally assessed to determine if there is evidence that one or more are likely to be pathogenic. This process is limited by existing information on the genetic sequence and the consequences of variation. Whilst many variants that are not pathogenic are prioritised there is a risk that a pathogenic variant could be missed due to the way the prioritisation models are built. Functional genomics can be used to better understand which genetic variants might be causing disease.
Case study – The transcriptome
Analysing the transcriptome (i.e. all RNA) can help to reveal pathogenic changes to gene transcription due to variation at the level of the DNA. Whilst RNA sequencing (RNAseq) is used in clinical diagnostics to confirm whether a suspected variant is pathogenic by analysing specific gene transcripts, the entire transcriptome (i.e. all transcripts within a sample) can be interrogated without prior knowledge of which variant may be pathogenic. This latter approach has only recently been employed in clinical research.
For example, one notable study used transcriptomic data to provide diagnoses for patients with a rare neuromuscular disease where WES or WGS had inconclusive results. Using RNASeq, scientists sequenced the transcriptome from affected muscles of 50 patients without a diagnosis. In this study the authors looked specifically at messenger RNA (mRNA), which is translated into proteins. The transcriptomic data enabled them to make a diagnosis where there was a pathogenic change to gene expression levels, as well as changes that would influence protein structure, function and/ or abundance.
The added value of transcriptomic analysis was evident as it provided 35% of patients with a diagnosis that would not have been reached by sequencing and analysing the genome alone. Changes to transcription can be predicted from the DNA sequence by computational tools, however these are limited in their accuracy and specificity and the study reports that several cases would not have received a diagnosis or would have been misdiagnosed based on these predictions.
Though this was a small scale study, the enhancement of diagnostic rate was reflected in a later study. They found that incorporating transcriptomic data alongside genomic data resulted in 4 of 12 patients with musculoskeletal pathology receiving a diagnosis when genomic interrogation on its own yielded inconclusive results. Others have used transcriptomic data to aid diagnosis including those with conditions such as mitochondrial disease.
So where are we now?
Whilst these studies focus on the use of transcriptomic analysis, others are exploring different biological datasets containing metabolomic, epigenomic or proteomic information, alone or in combination, to assist with rare disease diagnosis and to better understand functional consequences of genetic variation. As more information is gathered, it is likely that certain types of ‘omics data will be more or less useful for different types of clinical indications. For example, examining the metabolome could provide insights in the pathogenicity of genetic variants within genes involved in metabolic pathways.
Though the genome is the same in every cell of the body, the genes that are transcribed, their regulation and the protein and metabolite profiles change dramatically between different cell and tissue types, and over time. Access to the disease-relevant tissue for analysis can be a challenge, and early evidence suggests that easily obtainable samples such as blood are not good surrogates for hard to sample tissues such as the brain. Furthermore, as more tests are carried out to obtain information on different layers of biological activity, there will be increases in cost and turnaround time in interpreting and returning results.
Whilst functional genomic principles have made it to the clinic on a small scale – e.g. using targeted approaches to understand the function of specific suspected pathogenic variants – the use of multiple, genome wide ‘omics datasets is some way off. It is not yet clear whether these approaches are ready or suitable for clinical diagnostics, however, evidence is gathering around the utility of untargeted functional genomic approaches. Only a subset of patients with rare disease will benefit from these approaches so selecting those patients most likely to benefit will be necessary to optimise use of these approaches within healthcare.