Life Sciences – Genome Research - Bioimaging: Mining a large image repository to extract new biological knowledge about human gene function

SD ID: Mining a large image repository to extract new biological knowledge about human gene function

Organisations & contacts

Prof Jason Swedlow, University of Dundee, Euro-BioImaging

Dr Alvis Brazma and Dr Jan Ellenberg, EMBL, Euro-BioImaging

Dr Jean-Karim Hériché, EMBL

Mr Balaji Ramalingam, Mr Josh Moore and Dr Simon Li, University of Dundee, Open Microscopy Environment

 

OVERVIEW:  Image-based genome-scale RNAi and small molecule inhibitor screens generate a wealth of image data that remains unexploited after their original publication. The science demonstrator establishes the resources required to perform comprehensive machine learning analyses on these datasets, along with a validity test of the approach and the demonstration on how large cloud-based collection of published datasets can be reused for novel discovery.

SCIENTIFIC OBJECTIVES OF THE DEMONSTRATOR:

  • comprehensive machine learning analysis

  • dataset reuse

  • reusable infrastructure and analysis for generating value from published image data

IMPACT:

  • demonstrate the use of the infrastructure for users to run their own analysis via the cloud on publicly available image data sets

  • show how image data can be reused in a research context

  • showcase the results reuse by the community e.g. for searching images by similarity, to implement supervised machine-learning methods to mine the repository, etc.