HealthHack: the musical
Imagine an event generating multiple innovations in medical research in one weekend. Imagine an event giving spreadsheet-limited laboratory researchers a taste of what computers can really do. Imagine an event where the powers of coding and analysis are combined with powers of biology, genetics and medicine to convert dry or unobtainable data into ground-breaking medical insight. And imagine it all done in a spirit of frenzied teamwork and friendly competition.
This is HealthHack, a 48 hour hackfest in which medical researchers nucleate teams of brilliant, creative people to solve the data-processing problems faced by real researchers and real clinicians.
The weekend opens on Friday evening with the problem owners pitching their ideas, problem and proposed solution, to the assembled hackers, followed by an hour in which hackers can approach problem owners with their questions and seek additional details. Hackers then choose which problem they are going to work on, and the assembled teams plan their assault and commence work on their application.
Apart from going home (presumably) to sleep, the event runs all weekend. Sunday 4pm is down tools, and the teams then present their prototypes to everyone, with prizes for first place, second place, and spirit of HealthHack.
Below is a sampling of the teams and their creations:
Scott Ritchie (@sritchie73)
A Ph.D. student in the life sciences who works on Bioinformatics problems. He has a background in computer science, and data analysis. He is also interested in data visualisation.
Sean Fleming (@a_sean_fleming)
A developer with skills around .Net and SQL databases. Also dabbles in chef, nodejs, mongodb, and a few other technologies.
Irith Williams (@IrithWilliams)
A user experience designer/researcher who should be at home writing up her Masters. Manages the LinkedIn group ’Designing for Health in Australia’... come and talk about making healthcare better through design!
Iulian Stefanica (@stefanicai)
Frontend and backend (Java) developer who should be sleeping now! Getting used to playing with data visualization ... hackathons are great for that! Founder at www.researchyo.com.
James Walter (@jamespwalter)
Logistics Analyst at Australia Post. Specialises in Data Visualisation, Maps/GIS, and general Analytics.
Aaron McAleese (@kiwiessential)
Data science and analytics consultant - specialist in driving optimal solutions in organisational and marketing strategy through intelligent and creative data mining, statistical analysis and customer insight.
Fred Michna (@fredmichna)
I.T for resilient communities researcher. Maps, society, environment and participatory technology. about.me/fred.michna
Problem Owner Marguerite Evans-Galea (@MVEG001)
Biomedical researcher developing therapies and biomarkers for neurodegenerative disease. Advocate for science and research, and also the research workforce: examining issues including funding, career structure, peer review, gender equity - all fall into science policy and/or policy for science. Molecular biologist, author, educator and mentor.
To present data on successful NHMRC grant applications in an accessible, understandable manner.
Walter & Eliza Hall Institute (firstname.lastname@example.org)
Michael Walker PhD student, MSB-group, Pathology, MDHS, Uni of Melbourne (email@example.com)
Amateur Dev, Physiotherapist (firstname.lastname@example.org)
Software Engineer+Architect, Distributed Computing, Machine Learning (email@example.com) (firstname.lastname@example.org)
ThoughtWorks Australia, Open Knowledge Foundation (email@example.com)
Biologists are creating a wealth of data that reports not only the DNA sequence, but also the quantities of various epigenetic marks, along with other different factors bound to the DNA throughout the whole genome. This landslide of data is highly complex, and difficult for bench biologists to interrogate.
While there were systems, each with its own pros and cons, for viewing such data already available (e.g. Galaxy, UCSC Genome Browser, IGV, SeqMonk), none provided a holistic view of the data that allows optimal visualisation. A histone display was particularly important because our problem owner (M. Ritchie) was particularly interested in the epigenetic control of cell differentiation.
Epiview is an easy-to-use software package which allows the user to view many different layers of genetic and epigenetic information at once, and to easily switch between viewing a single region of one gene, to many genes, and even a whole chromosome.
Visualising the genome, and in particular its epigenetic profile, in a Circos plot allows biologists to interrogate the data through an intuitive interface. The implementation is web-based, so the user can simply upload their data (.BAM files) and specify the chromosomes, and which parts of the genome data to display.
Epiview is available from git repository (http://github.com/mritchie/epiview) and was developed using the Bam files from Encode project (http://genome.ucsc.edu/encode/). It is based on the R libraries RCircos, to generate the Circos plots, and Shiny, to be operable via the web. This is enhanced by D3.js to facilitate interactivity.