Research Platform Services Blog

by Scott Ritchie

Earlier this month we hosted a one day workshop at the University of Melbourne focused on open access publishing and tackling the reproducibility crisis currently plaguing the life sciences.

In the morning there was a series of talks about how we can go about making our own work more reproducible through open access publishing and open source tools. This was followed by an afternoon of workshops, one on the use of Galaxy on the NeCTAR Research Cloud for conducting reproducible research, and another on Authorea for writing open access articles.

After the speakers’ panel, we asked everyone who attended to come up to the whiteboard and vote on what key outcomes they would most like to see followed up on next. Overwhelmingly, the vote was in favour of more training in tools for open access research and publishing, with over 85% of the attendees voting for this outcome.

— A few of the attendees stayed behind with the panel members to discuss reproducible research. 

The workshop was put together by Nicholas Wong, and run in partnership with the Australian Bioinformatics Network and The Research Bazaar. Sponsorship was kindly provided by the Australian Bioinformatics Network, Illumina Australia, VLSCI and The GigaScience Journal.

by Scott Ritchie

Although nearly every researcher uses Wikipedia, very few contribute to it. Over the last few months we’ve been working to change that for researchers at the University of Melbourne: we recently ran a workshop teaching researchers how to make edits, and supported a local student group, COMBINE, who ran their first ever national Wikipedia edit-a-thon for computational biology postgraduate students.

The event was a massive success, with over 20 postgraduate and early career researchers attending the Melbourne event that #ResBaz helped coordinate. We were also joined by students from around the country via Skype, from various institutes in Canberra, Sydney, Brisbane, and Adelaide.

This event was kindly sponsored by Australian Bioinformatics Network. The Melbourne event was further supported by the University of Melbourne Student Union, and the Graduate Student Association, and run in partnership by The Research Bazaar with COMBINE, and the Bioinformatics Graduate Student Association at the University of Melbourne.

By Damien Irving.

At a typical Software Carpentry bootcamp, participants install the required software on their own laptop (i.e. the bash shell; Python, R or MATLAB; and Git). This solution works OK (we’ve run several bootcamps that way), however participants on older laptops often have trouble getting the software to work, and even if it does work it tends to look and behave a little differently depending on whether the operating system is Windows, Mac or Linux.  

Cloud computing and its uses for research; @DrClimate explains - http://t.co/GzAbkIPFr7 (via @ResPlat)

— Melbourne University (@unimelb)

September 25, 2014

Given these issues, we’re very excited to have just completed our first bootcamp on the NeCTAR Research Cloud. Using the cloud essentially killed two birds with one stone:

1. Every participant had a working and identical copy of the software, which meant they could focus on learning programming as opposed to wrangling with install issues and operating system discrepancies.
2. All Australian researchers have access to the NeCTAR Research Cloud, but most don’t even know it. The bootcamp was therefore a great awareness raising exercise.

#IPython notebook on the @projectnectar Research Cloud! @swcarpentry @ResPlat #ResBaz pic.twitter.com/xiaKoMUOXZ

— Damien Irving (@DrClimate)

September 18, 2014

Our solution to getting tools like the IPython Notebook and R Studio working in the cloud has generated a lot of interest in the Software Carpentry community, so in the coming weeks Tim Dettrick (our lead developer and all round computing guru) will write a post on the topic for the Software Carpentry blog.

If you missed out on attending the bootcamp (there were 49 people on the waiting list!), we have more coming up soon. Alternatively, you can check out the video recording of the bootcamp on the ResBaz YouTube channel.

Genomics researchers, much like cattle drovers in the outback, are increasingly having to wrangle very large datasets.  Managing herds of DNA data can be a painstakingly manual chore for researchers, yet this is where the automation of machines can make your research life easier. 

Ergo: we are pleased to announce a new data service aimed at helping biomedical researchers working with large datasets as produced by DNA/RNA sequencing machines.  We are calling this new data service the “Data Drover”[1] and it is powered by the MyTardis system[2] and Research Bazaar[3].

The problem: DNA sequencing machines produce large datasets (terabytes in days!).  One computer server sitting next to your sequencer is not enough (nor is it secure!). You need the ability to share datasets with fellow trusted researchers, as well as utilise new analyst tools to decode the genomes; and the ability to publish your data to cite in your forthcoming research paper.  

The solution:  The Research Platforms department at the University of Melbourne is now offering consultation through its “Data Drover” service to any research group requiring a purpose built system for managing the research data produced by your sequencing machines.  We are very pleased to be using the tried & tested MyTardis research data system as the engine which automatically droves the data from your sequencing machine into well organised and described datasets.

Some of the much loved research features of the MyTardis system include:

• Captures data off a wide variety of scientific instruments and stores it securely and privately online.

• Allows the sharing of data securely with fellow research team members.

• Quick thumbnail previews of experiments, images and metadata.

• Easy-to-use web interface that offers search and “tagging” functionalities.

• Shareable timestamped links to allow researcher outside of your team to consult on your data for declared periods of time.

• Verify, backup and redundancy of data in University housed datacentres.

• For good practice in research data management techniques.

• Ability to produce DOIs of datasets for citation in your research publications once ready for publishing.

• Plugins that allow preliminary processing, and visualising of data.

But wait there’s more!  Our Data Drover service also builds on the MyTardis engine to help your team develop the skills to analyse your data using the latest genome tools like:

• Galaxy Genomics workflow engine, so you can do DNA analysis which is repeatable with clearly published methods. See: an interview with a researcher using Galaxy.

• Rstudio data analysis and research graph generator.  Helping make sure you produce graphs and plots which have good statistical analysis and error bars! see: #DataCarpentry

• iPython Lab Notebook for doing big data analysis across multiple gene sequences. see: #SWCarpentry

In short, our new Data Drover service can help you muster your DNA data management and analysis in the most efficient way possible so you can forget about the technology and focus on the research!

How can I get involved in the Data Drover Service so I can plug my sequencing machine into MyTardis and get my team skilled-up in utilising genomics analysis tools to the best of their ability?

Please contact Dr. Andy Tseng (Lead Data Architect <andy.tseng@unimelb.edu.au > or David F. Flanders (Research Community Manager) <David.Flanders@UniMelb.edu.au>.

The Director of Research Platforms is Dr. Steven Manos <smanos@unimelb.edu.au>

Stop by our website to pick up a brochure describing our other compute and data services.

Get the latest news from our news feed: https://twitter.com/ResPlat ← if you have a Twitter account, you can also ask question live to our friendly staff waiting to help.

[1]= Data Drover is a sub-project within the VicNode (a federally funded project to provide storage solutions to suit a variety of Victorian research data storage needs) aiming to offer a simple integrated workflow between instruments and data storage.

[2]= MyTardis (http://mytardis.org) is an Open Source data management system purpose-built for managing large datasets produced by scientific instruments.  It was created at Monash University with the Australian Synchrotron in mind (see https://store.synchrotron.org.au/) but is quickly becoming the de facto multi-instrument system for managing scientific equipment throughout all labs in the University.  To know more about MyTardis and its community, please contact steve.androulakis@monash.edu

[3]= The Research Bazaar is our campaign, community and conference to help bring about the next generation of digital research skills: http://resbaz.tumblr.com/about

Image citation: images by Kate B. Dixon, reused through a creative commons license.  Thank you for sharing Kate, your pictures have helped make science more interesting!

By Steve Bennett

Last week saw the first test drive of our our new workshop, “Get your research collection online with Omeka”. Omeka is a simple, powerful web-based collection management tool, that does at least three things really well:

1. Manage a collection of images, videos, audio, or even just descriptions of things.
2. Turn a spreadsheet of data into a collection of things.
3. Share some or all of your collection with the world, through an attractive, themable interface.

It has great metadata management, it’s really easy to add new fields, and you can follow standards like Dublin Core if you’re into that kind of thing. And it can even serve as the platform for some of the most richly annotated digital humanities exhibitions I’ve ever seen.

Six librarians and archivists took the plunge. Rather than ludicrously teach metadata to librarians, I just pointed them in the direction of a few challenges. They worked in threes, learning how to create collections, upload images, import spreadsheets, change themes and more.

It was meant to be a 2 hour workshop, but 90 minutes in, they were done. Finished. Out of there. For people used to working with clunky, ancient collection management software, Omeka really is a breath of  fresh air.

Quote of the session: “This is like…really good…and…easy!”

Omeka is free, open source software. You can get access to a basic instance on omeka.net for free, or Research Platforms can host it for you on the NeCTAR Research Cloud on request.

By Paul Mignone

This week I had the fantastic opportunity to discuss the upcoming 3D printing showcase on Triple R independent radio. With presenters Dan Salmon and Vanessa Toholka, we discussed what’s happening at the showcase as well as some of the interesting items I managed to bring in. These items included the actual bronchial tree of an adult human, as well as Ned Kelly’s Death Mask!

Click here to listen to the interview.

Interested in attending, you can register here!

www.3dpshowcase2014.eventbrite.com 

Ned’s head baby, Ned’s head…

by Scott Ritchie

Last month, I had the pleasure of being invited by David Lovell from the Australian Bioinformatics Network to speak about Software Carpentry at the Education and Training breakout session at the International Conference on Bioinformatics (InCob) 2014. 

#Incob2014 showcasing the experience with @swcarpentry in Australia

— VickySchneider (@MVickySchneider)

August 1, 2014

This session featured a series of “lightning talks”, 8 minutes in length, from a number of speakers giving student, course coordinator, and tool trainer perspectives on education and training for Bioinformatics research tools.

During the session, everyone in the audience was told to take notes on what they identified as enablers or barriers to training, as well as any other insights they gained throughout the session. Once all the speakers had presented, there was a half-hour interactive discussion between the speakers and the session audience about common themes in these notes.

The next day, I met up with Vicky Schneider, the vice president of the Global Organisation for Bioinformatics Learning, Education, and Training (GOBLET), along with Andrew Lonsdale, Marek Cmero, and Harriet Dashnow who were at InCob to talk about their experience taking the Master of Science (Bioinformatics) at the University of Melbourne, and to promote COMBINE, a national student group for computational biology. 

Between us, we a had a productive discussion on the state of education and training in Bioinformatics, solidifying our thoughts from the previous day’s session. Vicky Schneider sent out a tweet later, with an image summing up our collective thoughts:

My first impression on Bioinformatics in AU training & education @ausbionet @BioplatformsAus @combine_au @sritchie73

— VickySchneider (@MVickySchneider)

August 5, 2014

While our discussion on education and training was limited in scope to the field of Bioinformatics, many of the insights we had are also applicable more broadly to education and training for research tools across all disciplines:

More than anything else, researchers want training
At least in the life sciences space, the resource researchers wanted most was tool training. Mark Crowe from QFAB shared the results of the BRAEMBL survey conducted in 2013 of bioinformatics needs. More than 60% of researchers surveyed said that their greatest need was additional training, compared to a meagre 5% who need access to additional compute power. 

Although this survey was restricted to the life sciences, our anecdotal evidence at the Research Bazaar suggests that this overwhelming need for training is reflected across all other research disciplines as well.

Translating training to research impact through collaboration
Although the value of tool training is unquestionable, one of the biggest barriers we face as trainers is justifying our involvement in training programs. This was common for everyone in the room, from student to tenured professor. Those we answer to, whether it be our supervisors, or the funding agency providing grant money, see our involvement as detracting from our research priorities.

So how do we go about translating our involvement in education and training into research impact for ourselves? The answer: collaboration. As the saying goes, a rising tide floats all ships, so by providing training to other researchers we open the door to collaborations, as those who attend can now ask, and answer, more interesting research questions with their data. Those who attend training are more likely to collaborate with each other and with the trainers.

Fragmented efforts to provide training
One of the key insights from the education and training session is that there are currently many groups with overlapping training efforts. GOBLET, CSIRO, QFAB, BRAEMBL and VLSCI all provide training in the life sciences, as do we here at the Research Bazaar. 

This was one of the main points of discussion with Vicky Schneider from GOBLET. We both saw a need to promote collaborations between these organisations to enhance our training efforts where they overlap. Pooling resources for shared efforts will also help us complement each other for the unique training needs each organisation provides.

Improving the quality of training
What I was most impressed to learn at the education and training session is the increasing focus from all organisations in improving their training. This is one of GOBLET’s primary functions: to act as a resource for bioinformatics training curriculum and best practices in teaching, and CSIRO have also run train-the-trainer events in the past.

Improving training quality is also one of our key goals at the Research Bazaar, and we will are holding a massive train-the-trainer conference next February to provide training in teaching pedagogy and best practices.

Students as enablers to training
One of the major barriers identified was a lack of awareness of available training (identified by BRAEMBL survey). Researchers in the life sciences are scattered across across many different labs, departments, and institutions, making it hard to effectively identify and advertise to those who want training.

This is where student groups, like COMBINE, come in. Student groups exist as a professional and social support group for those spread across disparate labs and institutions, so when appropriately targeted, can act as powerful enablers for training. They also bring together students with similar research interests and backgrounds, who thus have similar training needs. 

We’ve recognised this here at the Research Bazaar, and have adopted a model that has been incredibly successful: whenever we run a training event, we approach a student group to co-host it. The benefit of this is two-fold: we can effectively target a large number of students who want training, and can tailor our training to that particular group in discussions with the student group’s committee.

Getting students involved as trainers is also beneficial: both to the student and to the training. Those at the student level have a better understanding of the skills that they, and their cohort, lack, and are enthusiastic to gain teaching experience and improve their teaching ability. 

Symbiotic relationships between organisations who provide training, student groups, and student trainers is thus a powerful catalyst for enabling effective training for research tools. 

Session Slides
The slides for each of the speakers at the Education and Training session are available on slideshare, courtesy of David Lovell and Bruno Gaeta at the Australian Bioinformatics Network.