When Gamers Become Citizen Scientists in Bioinformatics

When Gamers Become Citizen Scientists in Bioinformatics

There is a continuing assumption that scientific advancement is fuelled by growing specialisation. The natural reaction to more complex problems has been to specialise, sharpen tools, and depend on more complex computational systems. This is more apparent than in bioinformatics, where the molecular level of understanding life can require exploration of spaces so large that even the most sophisticated algorithms can be unable to find stable solutions. However, in a twist to the story, some of these problems are no longer being solved in laboratories or data centres, but in the world we all know and seemingly take for granted, the video games.

This change was not initiated as a great redesign of scientific method. It was created as a bypass to a shortcoming. Protein folding, RNA structure design, and multiple sequence alignment are not only computationally intensive but also structurally complex. Algorithms work based on optimisation, which is directed by a set of rules and scoring functions. Biological systems are seldom so cooperative. They have anomalies, ambiguities and local pitfalls that tend to confuse even the most sophisticated heuristics.

Human cognition, on the contrary, developed in conditions when uncertainty was the rule. It is naturally efficient in identifying patterns, changing strategies, and making intuitive jumps without having to search all possibilities. What scientists started to investigate was whether such intelligence could be used, not individually, but on a large scale. The medium through which this idea could be tested was video games, which have the capability to involve millions of people in problem-solving activities.

The first and most successful example is Foldit, a game that breaks down the abstract problem of protein folding into a three-dimensional interactive puzzle. Proteins cannot work without assuming specific structures, but the problem of determining these structures based on the sequence of amino acids has been one of the most important in biology. In Foldit, players are manipulating virtual protein chains, rotating bonds and changing conformations to minimise energy.

What came out of this experiment was not only participation, but actual discovery. A study in Nature indicated that players could solve protein structures that had been difficult to solve using conventional computational techniques. More to the point, they came up with new strategies that were not directly coded into the system. Such strategies, after analysis, may be converted into new algorithmic methods. The game, in effect, was a place where human intuition created computational innovation.

The consequences were enhanced by subsequent studies on protein design. In another study in Nature, players were not only asked to fold existing proteins, but to make completely new ones. With an unfolded chain, they were requested to come up with sequences that would stabilise to viable structures. The results were striking. The laboratory synthesised dozens of player-designed proteins, many of which have been shown to fold into stable structures, including structural forms that have never been seen in nature. This was no longer a question of perfecting solutions known, but of increasing the known space of biological possibility.

The same trend was observed in the RNA research with the EteRNA platform. RNA molecules are simpler than proteins, but they have complex folding behaviours that are not easily predicted. In EteRNA, the player creates RNA sequences which need to be folded into a desired shape. With time, an international community of players created more efficient design strategies, which in certain tasks were more effective than the current computational models.

A study in Proceedings of the National Academy of Sciences showed that these player-derived strategies could be formalised into an algorithm that outperformed previous approaches. The direction of knowledge flow is what is especially interesting here. Human experimentation in a game-like setting produced new rules that were subsequently picked up by algorithms rather than humans. The line between the user and the researcher became less clear.

Although these platforms showed the potential of game-based science, they were still limited by the audience. It needed some interest in science or problem-solving. This model was actually transformed when it was incorporated into mass market video games, where the magnitude of participation was raised by several orders of magnitude.

Borderlands Science is one of the most ambitious applications of this concept. It is part of a commercially successful game that gives players puzzles that are analogous to several sequence alignment problems in microbiome research. In particular, the assignment will entail matching the pieces of 16S ribosomal RNA, which is one of the most important markers to determine the relationship between microbes and their ecological organisation.

On the side of the player, it is a visual and intuitive task. Coloured blocks should be grouped in such a way that they are as coherent as possible. However, every modification is associated with the matching of genetic sequences in a manner that can be used to affect subsequent biological interpretation. Multiple sequence alignment is infamously susceptible to local errors, especially in areas with deletions and insertions. These are some of the areas that algorithms fail at, generating alignments that need to be refined.

In a large-scale study published in Nature Biotechnology, millions of players together produced solutions that enhanced the quality of alignment in these challenging regions. These contributions, when combined, resulted in better representations of microbial phylogeny and improved the resolution of microbiome datasets. The scale itself is impressive. Millions of people were able to contribute to a dataset that would have taken years to compile using conventional means in a relatively short period.

The difference between this approach and efficiency is not merely efficiency, but the creation of a hybrid system. Machines give the starting point, working with huge amounts of data and producing first solutions. Through gameplay, humans perfect these outputs, especially in those areas where ambiguity is the rule. The outcome is a system that integrates the speed of computation and the flexibility of human thought.

This hybridisation is indicative of a wider change of approach. Conventional bioinformatics aims at finding the best solutions within a set of parameters. Game-based systems bring variety to the search process. The players search the problem space in different ways, by intuition, experimentation, and even aesthetic taste. This is not a weakness, but a source of innovation. It enables the system to get out of local optima and discover other solutions that would otherwise be concealed.

Important constraints, however, exist. Not every issue can be successfully converted into a gameplay without losing the necessary details. Such systems need to be carefully abstracted in their design so that the underlying science is not lost, but the interface is interesting. Also, the contribution of individual players may be sporadic, and it may be necessary to have strong aggregation techniques to derive meaningful signals out of noisy data.

These obstacles notwithstanding, the trend is hard to disregard. Video games are becoming distributed research platforms, able to mobilise masses of people in the service of scientific discovery. They confuse work and play, professional and amateur, laboratory and real life.

On a more fundamental level, this development challenges a re-evaluation of the production of knowledge. Science, which was traditionally considered a specialised activity, starts acquiring the features of a collective process. The wisdom is not just the result of solitary knowledge, but the combination of numerous minds, each bringing a grain of knowledge to a greater picture.

In this regard, the combination of gaming and bioinformatics is not a technical innovation only. It is a shift in perspective. It implies that with the distribution and proper channelling of intelligence, it can work on scales that were once unimaginable. And in this system, even the mere playing of a game is something more, a minor involvement in the continuing attempt to comprehend the basic frameworks of life.