Project Discovery

Sci-fi computer gamers help scientists analyse big data

Monday 15 May 17

More resources on the project

Why not make thousands of gamers do your (insurmountable) data analysis, instead of trying to do it yourself? It may sound crazy, but nevertheless, this was exactly what biotechnologist Emma Lundberg from KTH did – with great success.

What if you could gamify tedious data analysis and make dedicated computer players find patterns in complex biological pictures?

In her talk at this year’s Copenhagen Bioscience Conference on Data-Driven Biotechnology, Associate Professor Emma Lundberg from Royal Institute of Knowledge (KTH) in Sweden addressed this issue of big data and big data analysis.

The idea arose, because her group of scientists generate millions of pictures of how proteins are expressed in human cells. The pictures can help scientists understand the human body and all the proteins produced by its different cells, and relate this to health and disease.

But the photos take thousands of man hours to analyse. You can teach computers to make the analysis – called machine learning – but the computers need input in order to recognize patterns. Hence, humans still have to do the initial footwork.

Scientists got 70 working years for free

The idea of integrating a scientific question into computer games came from the Swiss start-up MMOS that together with Emma and her group at KTH, teamed up with the Icelandic company CCP games and the developers of the successful sci-fi game “Eve online”. Together they created “Project Discovery” – a mini-game for pattern recognition in microscope images, integrated into the game visuals, narrative and mechanics.

The game was designed so that players would get points each time they did a pattern recognition. Eventually, the most successful players could win a protective space suit for future quests. Emma Lundberg was even made into an avatar.

"This was mind blowing and more than we could have ever dreamed of"
Associate Professor Emma Lundberg, KTH in Sweden

“In the images, we often look for patterns, meaning images with similar structures, which represent the same types of proteins or cellular compartment where proteins are expressed. But the analyst doesn’t necessarily need to know this, he or she just needs to be good at pattern recognition,” Emma Lundberg says.

During the first year, after the project was launched, the players provided 27 million image classifications and spent nearly 70 working years in the mini-game.

“This was mind blowing and more than we could have ever dreamed of. One player even spend more than 16 hours in a row on our game,” she says.

Data needs careful sorting and evaluation

Project Discovery has now ended, and the scientists are currently analysing the results. Emma Lundberg emphasized that the collected data from the game needed careful evaluation and sorting, because some players might not have take the game too seriously and mostly wanted to score points. But with the right processing and filtering, the results were still quite impressive and useful, she argues.

The citizen science project was conducted on data from the Human Protein Atlas – a project run by Professor Mathias Uhlén, who is Scientific Director of The Novo Nordisk Foundation Center for Biosustainability, which was co-hosting the conference.

About the 2017 Copenhagen Bioscience Conference on Data-Driven Biotechnology

  • Approximately 180 scientists from industry and academia were gathered in Hillerød, Denmark, to discuss and present their work on Data-Driven Biotechnology in May 2017.
  • This year, the The Novo Nordisk Foundation Center for Biosustainability was co-hosting the conference. More on the 2017 Copenhagen Bioscience Conference on Data-Driven Biotechnology.
  • This conference was part of a series of conferences called Copenhagen Bioscience Conferences; Read more here:

News and filters

Get updated on news that match your filter.