Illumina: Difference between revisions
No edit summary |
No edit summary |
||
| (5 intermediate revisions by the same user not shown) | |||
| Line 1: | Line 1: | ||
==2016 proposal== | |||
[[ | [[Citizen Science for Cancer]] | ||
===earlier suggestions=== | |||
Proposed client: Lisa Murray | |||
Staff Scientist, Bioinformatics | |||
lmurray@illumina.com | |||
long-term contact Scott Oldham -(soldham@illumina.com) | |||
Suggestion: | |||
Genomics Mechanical Turk | |||
Analysis of genetic data (genomics) is full of classification tasks that are difficult to solve algorithmically, but which a human expert can often figure out from a quick glance at the data. Even novices have lots of insight to offer into these tasks by bringing novel perspectives and fresh ideas to these complex problems. Getting lots of people to solve thousands of instances of a problem for you lets you learn general principles from their proposed solutions, which can then be implemented algorithmically to improve. The trouble with getting people to try to solve a difficult genomics problem is that it’s often a bit complicated, and for most people, not very fun. But we can change that! There are several successful games created that help scientists solve tough computational problems, such as Genes in Space (Cancer Research UK) and Foldit (University of Washington). These programs help researchers identify broken genes in cancers and figure out properties of potential drug targets, both crucial problems that need solving to improve human health. | |||
Let’s make genomics fun! Create a Facebook or mobile game to solve a difficult genomics challenge—identifying tumor-causing cancer mutations. Fundamentally, identifying these mutations is a signal processing problem where the signals are sometimes weak and the noise is variable. One way to conceive the game would be to imagine a comparison of three pictures: a “reference” picture and two “test” pictures. A player would have to figure out which, if any, of the test pictures look like the reference picture. All the pictures would be abstract representations of DNA sequencing data. The key will be providing an engaging way of representing this data from raw sequencing input and creating an engaging game from this core concept that people will want to keep playing. We will provide the “problem data” and suggestions for potential translations into a game setting. | |||
==2014 project (prize winner)== | |||
* [[Evolve a Pet]] | |||
==earlier suggestions== | |||
* [[CheckMate]] | |||
* [[Genomes are fun!]] | |||
Latest revision as of 18:41, 6 November 2015
2016 proposal
earlier suggestions
Proposed client: Lisa Murray Staff Scientist, Bioinformatics lmurray@illumina.com
long-term contact Scott Oldham -(soldham@illumina.com)
Suggestion:
Genomics Mechanical Turk
Analysis of genetic data (genomics) is full of classification tasks that are difficult to solve algorithmically, but which a human expert can often figure out from a quick glance at the data. Even novices have lots of insight to offer into these tasks by bringing novel perspectives and fresh ideas to these complex problems. Getting lots of people to solve thousands of instances of a problem for you lets you learn general principles from their proposed solutions, which can then be implemented algorithmically to improve. The trouble with getting people to try to solve a difficult genomics problem is that it’s often a bit complicated, and for most people, not very fun. But we can change that! There are several successful games created that help scientists solve tough computational problems, such as Genes in Space (Cancer Research UK) and Foldit (University of Washington). These programs help researchers identify broken genes in cancers and figure out properties of potential drug targets, both crucial problems that need solving to improve human health.
Let’s make genomics fun! Create a Facebook or mobile game to solve a difficult genomics challenge—identifying tumor-causing cancer mutations. Fundamentally, identifying these mutations is a signal processing problem where the signals are sometimes weak and the noise is variable. One way to conceive the game would be to imagine a comparison of three pictures: a “reference” picture and two “test” pictures. A player would have to figure out which, if any, of the test pictures look like the reference picture. All the pictures would be abstract representations of DNA sequencing data. The key will be providing an engaging way of representing this data from raw sequencing input and creating an engaging game from this core concept that people will want to keep playing. We will provide the “problem data” and suggestions for potential translations into a game setting.
