2021-12-15 meeting at the squealing pig

[explodecomputer]

Gib meeting with Robert and Rafael during Boston visit

To discuss potential future work with OpenGWAS and Harvard CCB

Software development and integration into CRAN/bioconductor

Core packages that we control but need to be moved to a regulated source:

• https://github.com/mrcieu/twosamplemr/
• https://github.com/mrcieu/ieugwasr/
• https://github.com/mrcieu/gwasvcf/
• https://github.com/mrcieu/gwasglue/

Additional packages are core to the analytical ecosystem that we don't control and may be in github only. Also a question of creating a docker image release for core environments.

Ontology mapping analyses

Once we have ontologies mapped then there are a few potential use cases

1. Search function by navigating along the ontological hierarchy
2. Use the ontology as a prior for improving power of genetic association at child ontologies that share a parent ontology.
3. Use ontologies in automated multivariable MR - we want to identify the set of exposures that share instruments and automatically analyse them in MV-MR. However, that might accidentally include proxies for the outcome. Use ontologies to remove any proxies.
4. Traits can be clustered by navigating along the ontological hierarcy. It would be interesting to see how ontological clusterings correspond to the clusterings from embeddings e.g. in vectology.

Industry forum

Robert mentioned that he is in touch with many different companies who are interested in OpenGWAS. Start a forum in which we discuss future steps to develop OpenGWAS to be suited to some of the requirements in industry. This could be beneficial for its sustainability and cost recovery.

Improving costs by optimising data structure and storage

Costs are high largely because of the need for many Tb of fast data storage (currently in Oracle). Project to determine appropriate alternative methods for hosting and querying data e.g. storing in an appropriate format in a free bucket that is queried by the API and a download source.

Managing software engineers and the resource in general

Keen to discuss management strategies

[explodecomputer]

Present

Gib, Vince, Rafael, Robert

Ontology mapping project

Implement search by ontology using python library 'reasoner' against OWL format ontology db

Aim to write a paper that describes mapping process and compares ontologies against embeddings. Eg discovering and aligning across datasets, using the ontology as a tool in statistical modelling etc

Sustainability and cost recovery

1. We considered discovering or creating a database solution that can be hosted on a free cloud bucket. However the performance of elastic search comes from using fast SSD disk and a large set of indexes ie a lot of memory usage and therefore expensive. Discussed using Hail which would allow for greater interoperability with other genomics however to be performant it will need more development on indexing but also fast SSD in order to compete with elastic search. And development time esp r side.
2. Industry engagement for cost recovery could be necessary. Could try to make opengwas a non profit spin out that provides consultancy in return for dB cost recovery etc.
3. Grant funding eg elixir. Czi to be considered for the db question

Software development

• Put all packages into bioconductor - release versions are more reliable
• Consider openoauth2 for all users so that we can track usage more easily and throttle usage potentially also.