Norwich, United Kingdom
July 9, 2015
Chair of Computational Biology at the Sheffield Institute for Translational Medicine and Harvard Stem Cell Institute Winston Hide visited TGAC to give his seminar addressing the non-model organism community who have an interest in genomics and its application to the fundamental problems in computational biology and genomics. We asked Professor Winston to tell us more about the challenges faced in distributing knowledge in bioinformatics.
Chair of Computational Biology, Sheffield Institute for Translational Medicine and Harvard Stem Cell Institute Winston Hide came to TGAC to give his seminar on “Translating the genome: The next generation”. Addressing the non-model organism community who have an interest in genomics and its application to the fundamental problems in computational biology and genomics. We asked Professor Winston to tell us more about the challneges faced in distributing knowledge in bioinformatics.
John Hancock (TGAC), Professor Winston Hide, Vicky Schneider (GAC)
What key challenges does the rise of next-generation sequencing (NGS) pose for researchers?
The ability to generate sequence data at scale is upending conventional approaches to biological understanding. Classic one gene approaches are now put into the context of multiple data points that are simply beyond the ability of more mainstream scientists to address.
“The drive to sequence the human genome in the 1990s (sic) is a major point at which computational biology transformed into a research field. (sic) Lee Hood of the Institute for Systems Biology, contended then that biology was becoming a data science. By 1998, SmithKline Beecham had organized the largest corporate bioinformatics department, headed by computational linguist David Searls, and this investment brought new attention to the power and scale of computational biology."1
The fundamental issue is that researchers need to be able to handle the results of their own assays, and the NGS technology removes that for the majority of researchers - creating a need for interpretive skills, and for those that can interpret the data.
How important are successful training programmes, such as the ‘expert cadres’ programme that you helped set up, in overcoming these hurdles?
There is no single solution that can address the growing need for expertise. The ability to manipulate, assess, interpret and effectively synthesise results from NGS experiments using best practice is not something that a person can perform independently after a one-day or even week-long workshop. The skills required need practice, as well as a supportive community.
A smart approach is always to empower the end user - the person who is generating the experimental data that is providing the substrate for NGS analysis. To do that, however, requires that learners must congregate into a small communicative community that is embedded with expertise, and that they get access to this and each other over a significant period of time as they practice and gain skills. This approach is time-expensive but expertise yield-high - we have piloted it at Harvard and feel that it potentially scalable. It is mirrored by examples such as the CGAT programme at Oxford (Sims, David, Chris P. Ponting, and Andreas Heger."2
These cardre approaches are important, but they are challenging to scale - I believe in a multipronged approach that includes regular meetings, online training through MOOCs and shared materials, a vocal community of practice and open access to the physical forums available, coupled with mentorship by experts. That’s a tall order, but it’s the kind of community you can build from willing participants just about anywhere, given a few knowledgeable experts and the will to scale. We’ve done it in South Africa, across Africa and at Harvard. I see it here in UK in a few places, but am particularly excited to be putting this together with my colleagues at the University of Sheffield - a place where a community of practice is imperative.
NGS is also enabling genomics to play a greater role in medicine - could you outline the purpose of the upcoming training programme, Best Practice Delivery in Genome Medicine?
One of the reasons I moved to the UK was the big push by its healthcare system and government to embrace the potential of genomes and genomic medicine. However, the practice of that - in a nascent environment - is a real challenge. When I saw there was a call for development and provision of MSc training in Genome Medicine I jumped at the chance to make a bid. We are now forming a network of training centres across UK - with the singular purpose of empowering the healthcare professional to be able to deliver better care informed by genomics and genome practice.
The programme will give trainees the ability to have expertise in Quantitative, Technical. Functional, Clinical and Translational aspects of genome medicine. He or she would be able to work closely with the 100,000 Genomes Project and Microbial Genomes projects, to integrate genomic knowledge into healthcare and so make the most of genomic medicine as it applies to their work.
With the rising quantity of data, how crucial is it to ensure that there is a level of standardisation in research so that results are reproducible and comparable?
Any new technology goes through a disruptive phase and then a standards development phase. Although I’ve talked about commodotisation of genomics before, without standardisation we will be in the realm of ‘anecdotal importance’ where groups and companies make claims that are hard to reproduce - so we really need to go all out on reproducibility before we can seriously expect to be able to use genomics in everyday life.
What role is the Stem Cell Commons project playing in this regard?
The stem cell commons (stemcellcommons.org) is an important Harvard-based project that delivers provenance and reproducibility to stem cell research. It takes on the problems of having a consistent workflow for analysis that others can reproduce for the provision of consistent results from high-dimensional omics studies. We want to scale the access of this system to other domains and so I am busy working with the UK agencies and institutions to deliver a model that will allow for researchers to deposit data, workflow their research, and have others reproduce their research, in silico. With standardisation and reproducibility comes the opportunity to compare and search.
Are you optimistic about the future of open science?
There will be a lot of reluctance to perform open science until such a time as the benefit outweighs the perceived costs. Not sharing data before conclusions have been drawn from it is a historical tenet that is held by current physician-scientists, biomedical researchers and basic biologists alike. As the need to bring in many eyes to make a conclusion valid grows, the need to share data will outweigh the reluctance to be scooped. The leaders will be the communities that adopt open data sharing as a way of doing science, and as a way of discovery. I firmly believe in it, as evidenced by my contribution to the Advanced Medicines Partnership for Alzheimer’s 4, and for our active work in promoting companies to perform pre-competitive data sharing as a path to more rapidly making target choices for drug discovery.
Winston Hide's seminar was facilitated by Vicky Schneider, Head of the 361 Division at TGAC.
1) Computational Biology: Moving into the Future One Click at a Time Fogg CN, Kovats DE (2015) Computational Biology: Moving into the Future One Click at a Time. PLoS Comput Biol 11(6): e1004323. doi: 10.1371/journal.pcbi.1004323)
2) CGAT: a model for immersive personalized training in computational genomics." Briefings in functional genomics (2015): elv021.)
3) Hide, Winston (2013): Bioinformatics for the Public Eye
4) http://www.nih.gov/science/amp/alzheimers.htm
Species
