How did InterMine determine its FAIR milestones?

life-sciences-fair

(Cross-posted from my blog here)

At InterMine, a life sciences data integration platform, we’re working on a BBSRC grant to make data available through InterMine ‘FAIR’. What does this mean? Well, firstly  FAIR is an initiative to make dataFindable, Accessible, Interoperable and Reusable (I’ve written a lot more about this here).

Taken on its face this is a bit woolly – isn’t InterMine data already FAIR? You can find data (type some text in its general search box or perform a structured query), access it (click the web link), interoperate with it (run a live query on its API) and reuse it (hey the data’s there, download it). Well, one of the great things about FAIR is that it has specific principles and recommendations on how to make data findable, accessible, interoperable and reusable. These place a heavy emphasis on uniformity so that software can much more easily use and combine data across the countless distinct data sources hosted by different organizations across the planet.

So in applying for the grant, how did we propose to apply these recommendations to InterMine? Essentially, we performed a gap analysis between the 15 guiding principles documented in the original FAIR paper and InterMine’s current capabilities, coming up with a plan for how we would bridge this gap.

Let’s take the first findability and accessibility FAIR guiding principles as an example

F1. (meta)data are assigned a globally unique and 
persistent identifier

A1. (meta)data are retrievable by their identifier 
using a standardized communications protocol

One way to fulfil these principles, and something popular in the semantic web world,
is to make identifiers be URLs. So great, InterMine already has URLs that have a 1-to-1 mapping to biological data objects! Search for the gene MYH7 in HumanMine for instance, and the report page you get back has this URL (stripping away some non-essential tracking information).

http://www.humanmine.org/humanmine/report.do?id=1157771

Look at another biological object and that ID number will change, since this is the internal ID used to track objects within an InterMine database.

But there’s a problem here. These ID numbers are not persistent, as required by principle F1. When the data in an InterMine installation like HumanMine is updated, this is not done additively, but rather than entire database is rebuilt since data sources need to be integrated anew. And on this rebuild, MYH7 is no longer guaranteed to have the
internal InterMine ID 1157771. In fact, it’s very likely to be different.

So part of our proposal was to implement a resolution to this problem. For InterMine as a data integration platform rather than a primary data provider it’s a very complex topic, particularly as we’re generic and model driven (so in principle you could host something completely different like a company database in InterMine!). I won’t delve into the possible solutions too much here, but at the moment it looks like a tradeoff between trying to make our internal ID persistent (e.g. by maintaining the mapping to biological objects between database rebuilds) and trying to incorporate external IDs such as MYH7 directly into the InterMine URL as specified by the InterMine instance operator, something like

http://www.humanmine.org/humanmine/gene/MYH7

We’ll be reporting more on this in the future.

This was a fairly straightforward example. Some of the other principles, such as

I3. (meta)data include qualified references to other 
(meta)data

required more interpretation, and in our proposal we related actions broadly to the principles (i.e. whether they addressed one or more of findability, accessibility, etc.) rather than specific FAIR clauses.

However, we wrote our proposal some time ago. Things are moving rapidly and many of the original FAIR paper authors are working on the FAIR metrics initiative, which will measure FAIRness with programattic and quantitative tests. I think this is a great step and now something for anybody looking to FAIRify their data resource to look at closely. We’ll be looking to apply these metrics to our own work as we continue development.

 

Advertisements

Cambridge Science Festival 2018: A fruity crime of passion 🍏🍋🍊🍓

TL;DR: Science Festival was great & kids loved it. You can re-use our materials, here.

Longer version: Last weekend was InterMine’s very first year at Cambridge’s famous Science Festival, an event designed to enthuse younger people and adults alike with awe for science. We split our time across two locations,working at our home department, Genetics, on the Saturday, and at the Cambridge Guildhall on the Sunday.

Our theme was around open science, with an activity designed to reinforce the idea that shared data (and therefore more data from different sources) results in better science. For adults we had a couple of great posters about the importance of data sharing, designed by Julie and Rachel. The posters are available freely online for re-use under a CC0 licence.

The Story: A party is rudely interrupted

Meanwhile, for kids (and some adults too!) we had a crime-solving activity. In our scenario, a dastardly fruit villain had stolen the passionfruit in the midst of an otherwise enjoyable soirée. In their haste to flee, the culprit knocked over a tin of blue paint, leaving tracks behind, as well as injuring themselves and leaving DNA evidence behind as they jumped out the window. We had four fruity suspects:

suspects-sheet.png

Solving the crime

Step 1: footprints in the paint

In order to solve the crime using science, our young detectives were invited to examine the footprints left by the culprit:

Fruit tracks at the crime scene. Excuse the glare from the plastic!
Fruit tracks at the crime scene. Excuse the glare from the plastic!

It was usually pretty easy to rule out the apple, and after thinking a little more, the strawberry could be ruled out too, but the orange and the lemon both looked rather similar.

Step 2: Juice found at the scene

Since the devilish thief had hurt themselves, we had samples to analyse. Our criminal investigators took strips of litmus paper and carefully examined the evidence:

20180318_105143

Once again, the evidence wasn’t quite conclusive (and was very sticky). Still, it was fun! Let’s move on to the next bit of evidence…

Step 3: the skin

With sample fruits to compare, our enterprising criminologists got a step closer to the solution. Could the skin be from a lemon? Hmmm.

20180318_105151

Step 4: We have samples, so let’s sequence the DNA!

Okay, so you may have guessed that we didn’t sequence the DNA of the suspects ourselves – but thankfully the lab had four profiles for us to compare to and they managed to quickly provide a DNA fragment from the crime scene evidence, too. This fragment was far more conclusive than the others, pointing unequivocally to the shadiest character of the bunch – Lithium Lemon.

Step 5: Putting the puzzle pieces together, and sabotage!

As our sleuths solved each different activity, we gave them a puzzle piece. At this stage they had four pieces of the puzzle, but they were still missing a couple of critical bits: the two central pieces. It turns out there had been some CCTV footage – but it had been stolen! After looking around, our vigilant investigators discovered where the crime scene video had been hidden (under the table) and managed to put the entire story together. Once again, shown front and centre of the puzzle was our suspect, Lithium Lemon.

fruit-bowl

 

Wrap up

While the shady character wad hauled off in cuffs to the county jail, successful detectives were rewarded with candy, some awesome stickers,  and a handout that had a child-oriented activity sheet on one side, with a small copy of our open knowledge posters on the other side, for the slightly more grown-up folks.

What we learned

Our tables were generally very busy, and the kids seemed to have a great time examining the evidence and putting together the puzzle pieces one by one. I’m not sure how many of them quite perceived the data sharing theme, but some of the adults definitely did, and appreciated the posters as well.

I think one of the biggest surprises for use was how busy we all were! Genetics had a steady flow of people, but the Guildhall had even more. We haven’t heard numbers for this year yet, but in 2017 apparently there were around 3,000 people. What that meant in practical terms for us: Two tables with identical versions of the activity, two InterMine team members acting as detective wranglers at each table, and often two separate groups of people working through the activity simultaneously at each table. After several hours of this we were all ready for a nap! Next time, six staff might be better to allow people to have a breather.

We also learned to keep a good eye on our puzzles: Five puzzles left the office on Sunday morning but only four returned. Hopefully it’ll be cherished at someone’s house as memories of a great activity…. ?

Our materials are open!

Given that our activity was designed to advocate openly sharing your science, we’ve shared our materials online too, and you’re welcome to re-use them.

https://github.com/intermine/science-festival/

This includes:

  • The fruit images (lovingly created by Rachel’s daughter!)
  • Handouts
  • Posters
  • Guidance sheets and in-depth “sciencey details” about each activity.

If you do re-use them, we’d love to hear about it! You can email info@intermine.org, tweet @intermineorg, or even open an issue on the GitHub repository.

Finally, I’d like to thank Rachel again for all the work she put into designing this scenario. It was creative, exciting, and overall seemed to be a hit!

 

Google Summer of Code: Let the enquiries commence!

Last month we applied for InterMine to join Google Summer of Code (GSoC) as a mentor organisation, and we’re pleased to report that we have officially been accepted!

Students: Interested in working with us for GSoC?

Our GSoC site has a project ideas list and the student application guidance, which hopefully will answer most of your questions.

Want to learn more?

  • You can also read our GSoC blog posts from last year to learn more about how things went.
  • If you still have questions:
    • If the question is project-specific: email both listed mentors of the given project.
    • If the question is about GSoC in general, see the student manual.
    • We’ll be running a GSoC question and answer video call session where students can learn more about the specific projects. Updates about the exact date and time will go out on this blog, our mailing lists, and twitter.

We’ll look forward to hearing from you!

 

Community Spotlight: Correlating fly gene expression with FlyCAGE – Interview with @codingbash

For today’s blog we’ve interviewed Basheer Becerra, an undergrad bioinformatics researcher at the Illinois State University, about the application he’s created, FlyCAGE – (Correlation Analysis on Gene Expression). (Live Demo, GitHub)

Hi Basheer! We love hearing about new and interesting uses of InterMine. Can you give us a brief non-technical intro to FlyCAGE? What inspired you to make it?

FlyCAGE is a web application that allow users to search for genes in Drosophila melanogaster that follow a specific mRNA expression pattern. The user can either enter a known gene name to find other genes with similar expression profiles, or the user can enter a custom expression pattern based on experimental data to find genes that follow the pattern. FlyCAGE would be useful in identifying candidate genes involved in a given process and discovering regulatory interactions in genetic networks.

Tell us a little about yourself.

My name is Basheer Becerra, and I am currently an undergraduate junior at Illinois State University double majoring in Computer Science & Statistics and minoring in Biological Sciences. Programming is something I love doing, specifically web development and data science. What makes me even more excited is using programming and mathematics to answer difficult questions in biology! When I’m not on the computer, you will usually see me reading, training for upcoming marathons, or spending time with friends and family.

You’ve been a friendly presence in our Twitter feed for a while now. How did you hear about InterMine originally?

InterMine was originally introduced to me by my advisor, Dr. Nathan Mortimer, as a tool to help scientists quickly review information about any gene. While InterMine is a helpful tool for scientists to browse gene information, I’ve also realized that InterMine is incredibly useful for developers and data scientists. When my advisor and I came up with the idea of FlyCAGE, InterMine was chosen to be the best solution for retrieving data due to its data integration features and ease of use.

Can you tell us a bit about the technical implementation of FlyCAGE?

The technologies used to implement FlyCAGE includes Spring Framework (Java) for the web back-end, Thymeleaf for template resolving, and HTML/CSS/Bootstrap and JS/jQuery for front-end development. After the expression information is extracted from the entered gene or pattern, Pearson’s correlation is performed on every gene stored in FlyBase for Drosophila melanogaster. Genes with the highest pearson’s correlation coefficient relative to the input expression is returned to the user. InterMine plays a significant role in the operation of FlyCAGE since FlyMine is the only resource used to retrieve the gene information and mRNA expression data. With FlyMine’s modern HTTP API, only a single query is needed to retrieve all the necessary data for FlyCAGE to operate. Without FlyMine’s data integration, FlyCAGE would have to manually integrate data from several different data-sources such as FlyBase, FlyAtlas, BDGP, etc., which would slow down development significantly.

What are your future plans for FlyCAGE? Are you going to expand to other organisms apart from flies?

As far as the program logic, the plan is to include more complex analysis of expression data such as including other data features to help determine “gene similarity”, predict regulatory interactions and unknown gene functions, and explain subtle differences in gene pairs with correlated expression patterns. There has also been a lot of interest to expand CAGE towards other organisms such as plant genomes. With InterMine’s standardized API interface across its several resources, I predict that scaling the functionality of CAGE towards other organisms should be a relatively feasible task.

FlyCAGE is currently in-alpha and can be accessed by this link. However, the link is likely to change as FlyCAGE gets close to releasing. If you would like to stay up-to-date with FlyCAGE or if you would like to help us with usability tests to improve FlyCAGE, please enter your email with this Google Form. If you’ve already looked at FlyCAGE and would like to send some feedback, send me a quick note at bbecer2@ilstu.edu. Any help is appreciated!

‘Twas the week before Christmas… [aka InterMine availability over the holiday period]

… and all through the lab, not an organism was stirring, not even a… crab?*

Emails and support: Just a quick blog reminder that the office will be pretty empty from around now until the second of January, so don’t be surprised if we take a while to reply to messages. Some of us may be in the office or working from home, but it’s pretty patchy over the holiday season, and I don’t think any of us will be answering emails between the 23rd and 26th of December, nor on the first of January.

Developer calls: There is no developer call this week (normally it would be scheduled for Thursday the 21st). I’m not sure at this point, but the call on the 4th of January may (or may not) be cancelled as well.

Be good, have fun, and we’ll see you next year!

*I can only apologise for the terrible rhyme. Apparently nothing rhymes with “Cambridge” except “drainage” (and even then it’s somewhat weak), so I tried “uni” and that only rhymed with “loony” and “goonie”. Finding rhymes for the word “office” was no better.

Looking ahead: InterMine+Google Summer of Code 2018. Could you be a mentor?

2017 is coming to an end, and I have to say it’s been a fabulous one! I’ll probably post a “cool things InterMine did this year” round-up in a week or two – but in the meantime, here’s my final Google Summer of Code blog for you all!  We’ll cover the InterMine swag just sent out across the globe, as well as plans for next year – and how you can help out.

Thank-you gifts for mentors and students

Last week, we posted care packages to all our GSoC mentors and summer students, in the form of t-shirts, stickers, and pens. The postal-service-wrinkled shirt shown above is the women’s fit shirt printed on black; unisex shirts are a slightly lighter grey colour. If you filled out the swag survey when it was sent to you, your gift should be with you soon! Tweet us your images of the items in use for extra InterMine Cool Points 😎.

GSoC 2018 – call for project ideas and mentors!

Early 2017, we put together an ideas list for GSoC projects – InterMine’s projects are numbers 3 to 9. If you want to get more of an idea what it’s like to apply, (or be a mentor), read our application guidance from last year.

Do you have a nifty idea, or an InterMine itch you’d like to scratch?

Please share it with us! Add it to our 2018 Google Summer of Code ideas list, or if you need to sound things out and discuss them a little bit, comment on the GitHub issue, or email the dev list. You can even propose several ideas, if you like! Please add all ideas by the end of 14th of December (end of this week).

Would you like to try mentoring?

Fancy a chance to earn some nifty exclusive swag like pictured above? Add your name as a possible mentor to an existing idea (or your own new idea). You can always drop us a line if you want to discuss things first. We like projects to have more than one mentor if possible.

Maybe you’re a student thinking of GSoC?

Awesome! If you have your own InterMine project idea (whether it’s brand new or you’ve already started it), or if one of the ideas on our ideas list lights your fire, it’s not too early to start talking with potential mentors about it. The application guidance we mentioned above would be a good read, too.

 

 

#OpenConCam: Where open (science | access | source | data) meet.

What is OpenCon?

OpenCon is a yearly event designed to bring together people who are dedicated to open in all its incarnations. It’s in such high demand, the only way to get in is by application, and most attendees are provided with scholarships to help with travel/accommodation costs.

We weren’t able to attend the international event, but thankfully there was a great satellite event running in Cambridge – OpenConCam.

OpenConCam was in itself a day filled with memorable talks and worthwhile collaborations, including:

PeerJ – (Sierra Williams)

PeerJ is an open access journal which focuses on methodological rigour  when publishing, rather than preferring groundbreaking new science – something particularly important for early career researchers. One of my favourite points from her talk was when she demonstrated the checklist that PeerJ uses to help authors disseminate their content effectively:

Open access in developing nations (Tapoka Mkandawire)

Many of us know from personal experience that accessing scientific publications even in wealthier western countries can be controversially difficult, so it’s hard to imagine how much more difficult this must be in developing countries. Thankfully, there are initiatives such as Africa Information Highway, Eifl, and Hinari which aim to make data and publications more accessible. She also discussed the cultural concept of ubuntu – sharing and caring for each other as a concept that works hand-in-hand with the open* movement.

Bullied into Bad Science (Laurent Gatto)

Bullied Into Bad Science is a campaign to help early career researchers who may be under pressure to omit or tweak their scientific results in order to gain a desired outcome or exciting publication. Laurent was clearly passionate about this subject: Sometimes the system pressures mean that successful academics are not necessarily good scientists – and things really shouldn’t be this way.

Queen B

This session was frantic! The basic premise was that the room divided into groups of 4, nominated a “queen bee” who presented a problem (in one minute), and then the group broke up and discussed possible solutions with others in the room for three minutes, reporting back over the span of two minutes. Lather, rinse, repeat until all members in a group have been queen bees. Topics I recall discussing included getting humanities more involved in open science, open source code in science, how to inspire people to publish in journals with strict open policies when they could go for a less principled journal more easily, and how to sell open* to the disinterested.

Hitting a moving target in Open Access advocacy  (Danny Kingsley)

Danny shared something dear to our hearts: Getting others involved in open. While she was specifically referring to open access, most points could easily be applied to open science, data, and source too. Her focus was on figuring out how to get the most “bang for buck” – that is, find and influence people who will pay off the most for the least effort.

Undergrads, for example, aren’t great targets as they mostly don’t continue in academia, but PIs, and government bodies may be more useful, because they have much more influence if they’re sold on open access. Similarly, sometimes it makes more sense to influence decision makers and get them to evangelise for you, if you don’t have enough authority to impress people. Make sensible decisions, and don’t run up against brick walls repeatedly if it isn’t paying off!

Focus Groups

After lunch, we had an unconference-style set of sessions, where everyone nominated topics they were interested in, and added stars beside ideas they themselves were interested in attending. The resulting sessions were:

  • Self-care in Open: Many of us volunteer time outside a normal 9-5 job to help promote open, and the environment can be discouraging or rough sometimes – not everyone is as keep on open as we are! Suggestions presented by Kirstie Whitaker included working with micro-ambitions (turning your work into small, achievable chunks rather than trying to conquer everything), and thinking of success as a spectrum. A small win is still a win!
  • Open + inclusive: Laurent Gatto pointed out in a blog post earlier this year that the Open movements aren’t always as…. open as they should be. Sometimes Open Science can fall down in the same places less open science falls down – not making sure to have a decent balance of ethnicities, genders, sexual orientation, etc. Can we do better?

  • Open source code in science: If you’re an InterMiner, you’re probably already pretty keen on open source scientific software and can see the benefit of it – but not everyone does. Many, many papers that use code to produce their scientific results don’t expose that code. But if the code isn’t in the paper, or linked to it openly in some way… how was it peer reviewed? If the code is wrong, so is the science it produces. I proposed this discussion topic, and really enjoyed perspectives from my team mates. Some of the ideas generated included:
    • Share dummy data to run your code on, if the data are proprietary or there are privacy issues.
    • Try to encourage journals to have software availability statements
    • Encouraging researchers to share their code, even if it’s only a few lines. After all, if you’ve written 6 lines of code to configure an R plot, whilst it might seem insignificant – that’s actually really easy to peer review and correct mistakes! By comparison, bigger software packages can be hundreds, thousands, or even millions of lines of code. The thought of trying to review that (beyond reviewing quality metrics like testing, documentation, and commenting) makes me a bit scared.
  • Open in the humanities: This is a fascinating subject, and I don’t think many (any?) of the audience members were in the humanities. We raised a lot of questions about the shape of humanities data.

Opening the lab door (Christie Bahlai)

After the focus groups, Christie Bahlai skyped in to talk about running an open lab. She shared some of the different types of pushback against open science:

  • Those who consider themselves too busy to share
  • People who have been pushed from ‘busy’ status to actively hostile against open science, perhaps when they were asked to participate further and didn’t wish to
  • The worried –  people who have legitimate concerns about open science (I’m sure I’m not the only person who doesn’t really believe in “anonymised personal data”).
  • The unheard – those who are disadvantaged and marginalised already worry that practising open will marginalise them further. How can we protect these people?

She also talked about getting people involved in open as early as possible, including introductions to open as part of the undergrad curriculum:

A few more of her tips:

  • Get students’ feet wet in open science by slowly introducing them to the concepts using examples in their own fields – examples they’ll care about.
  • Share your lab policies openly and don’t tolerate the “brilliant jerk” – at the end of the day no matter how productive they are, they’re still jerks.
  • Keep science a kind place. Show others that you too can fail publicly, and fail often.
  • Share your lesson plans openly, too! Christie’s “Reproducible quantitative methods” curriculum is designed to provide a good introduction to open, reproducible data wrangling using R and GitHub.

The open source investigation revolution (Eliot Higgins)

This talk was an out-of-the-blue surprise. Rather than focusing on academia like most of the previous talks, Eliot shared how open videos, photos, and “facts” on the web can be verified for journalism. If you’ve heard of doxxing, you’ll know a bit about the techniques Eliot described, using social media, satellite imagery, and other online tools to track people who don’t want to be tracked – but this time, for Good. He described how some of the white supremacist rally leaders were identified, as well as verifying missile attacks in Syria – including who perpetrated them and who was lying about it.

This talk stilled twitter’s usually vibrant #OpenConCam discussions to a halt, probably due to the riot of emotions it induced in most of the participants. We’d been shown highly disturbing images, felt fear wondering how these techniques could be misused, and we awed by the massive importance of what we’re seeing, no matter how awful it was. I’m sure I wasn’t the only person torn between wishing I’d never seen it and knowing that I had to watch it, because burying our heads in the sand isn’t an option either.

Wrap-up

OpenCon 2018 hasn’t been announced yet, but this year, all around the world, there are still satellite events like the one I attended. If you haven’t attended a conference about working openly before, this is a great way to get a taste – or if you’re a die-hard enthusiast, you’ll get the chance to meet like-minded individuals and be inspired!