dnil
commented
2 months ago Hi Pär! Good to hear, and extra hands and brains are always much needed! We welcome plugins for external tools; chanjo is a reasonable template, but perhaps matchmaker or phenopacket will be more what you are looking for. We will certainly be available for for discussions and review, but I just want to repeat a couple of points to make sure you don't waste your effort.
- If the external tool is to do variant analysis for a case, it makes little sense to first run your own pipe, then import a few limited vars to scout select and export/api query to your external tool and get results back. The ideal workflow would be the external tool is fed raw data, does selection without loosing variants as potential false negatives to prior pipe and exports variants of interest for manual inspection in scout.
- Not taking notice of the tools in question, there has several times been discussion of tools that are not necessarily legally allowed for a basic clinical case, only specifically and carefully consented research cases. For such, we should limit visibility to cases where it is allowed.
parlar
vwirta
In Umeå, Scout has been selected as the primary tool for genetic variant interpretation and as a repository/database for variant information. However, there has been a request to also have the option to also use additional third-party interpretation tools, such as Qiagen QCI, Genoox Franklin, Roche Navify Mutation Profiler. This may prove useful in tricky cases. To facilitate this, we need a way to transfer data from Scout to such tools.
Scout provides a comprehensive environment for variant interpretation. It is our understanding that the Scout development team prefers to keep the core system streamlined, suggesting that additional functionality should be implemented as “extensions” rather than directly into Scout itself.
The planned feature is quite simple. It should enable selection of all, or a subset of genetic variants linked to a case, add additional metadata required by the third-party tool, and facilitate the transfer of the data and metadata to the third-party tool. Maybe in the future functionality for retrieval of results from third-party tools could be added but that will come later.
The problem is only how to do it. We have no previous experience of developing Scout features and extensions and therefore do not fully understand of the limitations of the Scout extension system. We are however very happy that we have been able to hire a consultant, Fredrik Silvearv, to work on this. (Fredrik has a background in quantum mechanics/materials science and is a good Python programmer.)
I was hoping that we might get some suggestions for a good way to implement this from the Scout dev team. We had a look at chanjo, which seems to be quite integrated into Scout itself despite being an extension. Would chanjo be a good example to follow or is that not how it would be done today?
Would it somehow be possible to put views into an extension, to avoid modifying the Scout server code too much. Or is this not possible? Or will it have to be done as a separate application, integrated via an extension. It's a small feature, would like to keep it a simple as possible.