I\u2019m a clinical bioinformatician working in genomics in NHS Wales. As a team, we provide genomic diagnoses to patients, but I personally work in the area of research and development. I take research from academia and apply it to a clinical setting, as well as working with academic researchers to provide access to NHS data. In a rather self-serving manner, I\u2019d like to use this blog to convince you why my job is important, why it\u2019s hard, and what people are doing to make it easier.<\/p>\n
In April 2003, the landscape of data in the field of genomics changed forever with the official release of the \u201cfirst human genome sequence\u201d. It had been 13 years in the making, costing ~$2.9 billion, and using expertise from around the globe, but finally, the \u201cfull\u201d DNA sequence of a human was available. In reality, this was just phase one, and since then, numerous iterations of the human genome have been released with improvements on the accuracy of the data. The most recent groundbreaking release came in 2022 from a team claiming that they really had completed it this time, with the full \u201cTelomere to Telomere\u201d reference genome. However, that groundbreaking first human genome in 2003 paved the way to make DNA sequencing, and the wider scientific field known as genomics, accessible to your everyday researcher.<\/p>\n
But why do we care about the sequence of DNA? Why was this such a big deal? DNA is the building block for all life. A structure consisting of four different molecules, known as nucleotides, wrapped up into the famous double helix, the order of which determines everything about that living thing. The colour of your eyes \u2013 that’s determined by your DNA, your blood type \u2013 that\u2019s DNA, how well a plant will cope with drought \u2013 DNA again, a banana being yellow \u2013 yep, DNA. In humans, approximately half of DNA is inherited from the mother, and the other half is inherited from the father. This is why people in families often share traits. Changes to that sequence of nucleotides means changes to the living thing.<\/p>\n
Changes can be passed on through family lines, known as inheritance, or can happen sporadically, known as de novo mutations. Sometimes this is beneficial, a great example being the discovery of a section of DNA, known as a gene, in plants which makes them tolerant of drought. Sometimes those DNA changes are not beneficial, with estimates suggesting that there are between 5,000 and 10,000 diseases in humans caused by changes to the DNA of a single gene, known as monogenic disorders.<\/p>\n
So of course, researchers immediately saw the potential of DNA sequencing to answer all of their questions! Suddenly everyone wanted their species sequenced, or the RNA from that species sequenced (telling us what genes are on and off), or the sequence from an entire population of individuals (telling us something about the evolution of that population). And with this, technology developed at a rapid pace to keep up with the demand and to provide an affordable solution. The field is now largely dominated by \u201cNext Generation short read sequencing\u201d primarily (but not exclusively) from Illumina, but recent years have seen the introduction of competing \u201cThird Generation\u201d or \u201cLong read\u201d platforms such as those from Pacific Biosciences and Oxford Nanopore Technologies. And with these affordable platforms comes a new challenge for researchers. What to do with all this data?<\/p>\n
Enter me, the bioinformatician! This specialty of science brings together biology and computational science, aiming to write code to process the large amounts of data efficiently and accurately, providing an \u201cend result\u201d with biological meaning for the researcher. A typical DNA sequencing run can produce several terabytes of data, and therefore innovative ways of processing this data are needed, as well as innovative ways to store this rather large resource in the long term!<\/p>\n
Within a clinical setting, DNA sequencing has opened up testing for patients with suspected monogenic disorders. Since the early 1900s, when Sir Archibald Garrod recognised that alkaptonuria, a rare metabolic disorder, was inherited in families passing from parent to child. Since then, diseases have been steadily linked to changes in DNA, such as in 1959 when Down\u2019s Syndrome was determined to be caused by an entirely extra copy of chromosome 21 (a very large chunk of DNA), and in 1983 when Huntington\u2019s Disease was linked to chromosome 4. However, methods for interrogating DNA to determine if disease causing changes have occurred were typically slow and laborious, looking at entire chromosomal structures as opposed to the individual nucleotides.<\/p>\n
DNA sequencing changed this. Initially in the 1970s, the advent of Sanger sequencing allowed for determining the sequence of, and therefore the changes in, short fragments of DNA up to about 1000 nucleotides. And then that wonderous full human genome was published, the methods for sequencing at a grand scale become more widely accessible, and next generation DNA sequencing became a method for determining the cause of an ever-increasing number of monogenic disorders. This started small with \u201cpanels\u201d of disease linked genes initially but quickly expanded into all of the bits of DNA responsible for genes (the exome) and eventually the entire genome sequence. Now in Wales, whole exome and whole genome sequencing within the NHS is carried out on several hundred patients a month, with a much larger number of cancer patients also undergoing DNA sequencing of tumours to determine appropriate treatment plans. This is because changes in the DNA don\u2019t just happen when the DNA is inherited, de novo mutations can occur over time. If these mutations happens in the sections of DNA responsible for its own maintenance, uncontrolled growth of cells can occur, which is what causes cancer.<\/p>\n
So researchers want the DNA sequence of everything, and in particular researchers of human disease really want the DNA sequence of patients. Better understanding of the DNA changes in patients could help researchers better understand the changes in DNA seen in the population, this leads to new regions being identified as disease causing giving more diagnoses to patients, it may identify targets for treatment and drug development, or identify eligibility for a clinical trial.
\nAnd the NHS is readily sequencing patients in order to provide a diagnosis. PERFECT! The NHS could just give all that data to the researchers, job done! No? Why not? DNA sequencing is a clinical test the same as any other, and therefore we\u2019re bound by laws and governance on sharing that data with others. Researchers aren\u2019t able to access blood test results or MRI scan results for example, unless they have a really good reason, and the patient says that they\u2019re happy with that! Processes are in place to make this happen, but it\u2019s not as straightforward as we\u2019d like it to be in order to share this wealth of knowledge with researchers.<\/p>\n
And it goes both ways! Academic researchers are doing fantastic work to push the boundaries of genomics techniques and technologies. The long read sequencing platforms and Telomere-to-Telomere reference genome are two great examples which I\u2019ve already mentioned. These were both born out of academic research. In order for the NHS to better provide diagnoses to patients, keeping up with the increasing demand and the increasing number of genomic diseases, we need to be readily testing and adopting new methodologies in the field of genomics. For this to be successful, this requires close collaboration between academics and clinical teams, with readily sharable data between the parties being key.<\/p>\n
Consent is often seen as the largest barrier. Unless the patient says they\u2019re happy for researchers to access their data, we can\u2019t provide access to it. It\u2019s as simple as that. You might ask, \u201cwell, why don\u2019t you just get consent from patients?\u201d. This is more complex than you might imagine. Often researchers want access to patient data that was collected already, something known as retrospective data. These patients have already been tested, already had their results, and are sometimes \u201cout of the system\u201d. Challenges lie in our ability to re-contact patients and then ask for their consent for research. This takes time which genetic counsellors and clinicians do not have and even being able to make that contact might not happen due to relocations or patients passing away.<\/p>\n
But ok, so let\u2019s say a researcher comes along with a study on a metabolic disorder, we manage to contact all of the patients with previous genomic testing for that disorder and get them to consent to research for that study. It was hard work, but job done. Another researcher hears about this and comes along with another study. It\u2019s the same cohort of patients. So now we need to contact them all again? That\u2019s just not feasible! Proposals have been made to add a broad research consent into the initial testing carried out for patients in genomics, which would cover all future research. However, this is difficult to do, adds a considerable amount of time to each appointment, and requires robust mechanisms for tracking who is consented and who isn\u2019t. This also doesn\u2019t solve the problem for retrospective patients who have already been tested and haven\u2019t given that consent. This problem of consent is one that is discussed often and at great length both within the NHS and academia, and several teams are trying to find solutions that work for everyone. The blog post from Rhys discusses the consent minefield in much more detail.<\/p>\n
As someone who works for the NHS, I\u2019m trying to test new technologies and methodologies, which sits within a grey area for consent. Under the Royal College of Pathologists guidelines, retrospective patient samples can be used for internal development of services. Determining if a project is a service development or a research project can sometimes be tricky. Often, a pilot study is needed as a proof of principle prior to a full service development. Guidelines, processes, and governance for these projects can be challenging, with individuals often disagreeing on the level of consent needed.<\/p>\n
Limitations of using data for only service development also exist. To prevent new results being found about a patient, which they perhaps didn\u2019t consent to or didn\u2019t want to know, clinical investigations are limited to look at what we already know. For example, Test A found a change on chromosome 21, the new Test B also found that change on chromosome 21. But what this is missing is that Test B also found three other changes that were missed by the first test as it\u2019s much more sensitive. Due to limitations around these incidental findings, we\u2019re just not allowed to look at this data.<\/p>\n
Ethical approval is also key. Researchers can\u2019t just ask for data because they feel like it. There has to be a sound scientific reason for the work ensuring that ethical guidelines are being followed. The Research Ethics Committee provides a process for this via IRAS applications which are then approved (or not) for research. However, this process is incredibly time consuming and long winded, often leading to significant delays in progressing research.<\/p>\n
Let\u2019s say this is an ideal world. We\u2019ve got consent, we\u2019ve got ethical approval. Now all that is left to do is transfer the sequencing data to the researcher. Simple right? DNA sequencing, and the subsequent computational processing of this data, produces many different file types of many different sizes. Education is needed so that researchers can better understand what they really want, and which subset of that data is the most beneficial for their research question. The transfer of the data itself is not straightforward. It\u2019s not practical, or clinically safe, to copy two terabytes of data to an external hard drive and send this in the post, hoping that the right person receives it and that it doesn\u2019t get damaged on the way. Instead, structured processes are needed to make this streamlined and safe, such as implementation of the specially designed procedures like the GA4GH htsget protocols as an example.<\/p>\n
When following the laws and guidance around using genomic data in research, one thing is pretty much always consistent. We must not be able to identify the patient from the data. But some will argue that genomic data itself is inheritably identifiable, especially when working with rare disease patients. If only one person in Wales has that genetic condition, and we\u2019ve got data that shows this condition, it\u2019s pretty easy to work out where the data has come from. Although in reality, this doesn\u2019t really happen, the risk itself prevents genomic data being used in research. Rob\u2019s blog post talks about the frustration and challenges in this space.<\/p>\n
A large proportion of what I\u2019ve discussed are the barriers to academic researchers accessing clinical data. However, a significant blocker for implementation of research in a clinical setting is funding. Academic researchers have (arguably not) easy access to funding streams for producing novel data. However, funding to take these findings and test them in a clinical setting is limited. There is nothing novel in the science, but what we really want to find out is if a technique is cost effective, functional and accurate in a clinical setting, which has much stricter criteria than academia. Funding streams for this work in genomics are more limited, and the surplus budget for this work doesn\u2019t exist in the NHS, leaving a gap in resources that ultimately slows or even entirely prevents work from happening. And that funding doesn\u2019t just have to account for the cost of equipment or the technology. Staff time is one of the most precious resources in the NHS, and surplus time of clinical staff is not readily available to carry out this work. Personally, this gripe is the biggest blocker I face in really being able to make a difference for patients.<\/p>\n
I don\u2019t want to continue too much down the path of barriers and challenges and problems. They exist, they don\u2019t make life easy, but researchers and clinical teams are tenacious, inventive, and determined. We find ways to make it work, following the laws and guidelines as we do it. And there are some fantastic success stories within Wales and the UK of what can be done when research and clinical genomics meet.<\/p>\n
The 100,000 genomes project run by Genomics England sat very much on the bridge between research and academia. Patients were recruited, their entire genomic DNA was sequenced, reanalysed in a research setting and findings were reported back to patients. This data was made available to researchers within a secure environment and still provides a valuable source of information eight years after the study closed. Patients in Wales saw the benefits of this work, also being recruited to the study, and receiving genetic diagnoses that had been missed in previous testing.<\/p>\n
Efforts are being made across several teams to make genomics data from within NHS Wales available to researchers. Working closely with teams such as SAIL Databank, who operate an accredited environment for access to sensitive data without consent, and the Wales Cancer Biobank, who provide access to samples and data from cancer patients who have consented to research. These pathways and procedures are taking shape to bridge the gap between clinical data and academia.<\/p>\n
Taking research and innovation ideas, and then implementing these into genomics services to benefit patients is something that the All Wales Medical Genomics Service takes pride in. Two notable projects in this space are WINGS and QuicDNA. The Wales Infants & Children’s Genome Service (WINGS), who were the first to introduce rapid whole genome sequencing for critically unwell new borns and children in the UK, was based on research and pilot studies carried out in the USA. The model was then taken and implemented into a service for Wales. QuicDNA took research into circulating tumour DNA, small bits of DNA which break off of tumours and circulate in the blood, in lung cancer. This was used to test the efficacy of genomic testing on blood, a far less invasive procedure than testing lung biopsy. This has now been rolled out into a service available to Welsh patients and is looking to be further expanded into other cancers. And it doesn\u2019t stop there, work is on-going in AWMGS to test new technologies, such as the long read sequencing platforms already mentioned, to provide improvements in the efficiency of genomic testing.<\/p>\n
Hopefully I\u2019ve convinced you of the value of close collaborative relationships between academia and the clinic for genomics research and shown that despite the challenges, fantastic innovations are out there. Why is it important that we make this all easier? Academic research in genomics feeds directly into improvements in clinical testing which will help to get patients more accurate results, faster. And NHS data feeds directly into researchers making discoveries on genomic diseases and treatments, which the NHS can then pick back up, implement and help to get patients more results or treatments. A raft of people from both clinical and academic settings are working tirelessly to improve this. Because, at the end of the day, all anyone wants is for those patients impacted by genomic conditions to get the diagnoses, treatment and support that they deserve.<\/p>\n