Updated Knowledgebase

[PabloBotas]

Hi! Do you have pipelines to update the KB?

Thanks!

[jimhavrilla]

If you're asking if there is a cron job or CI or something that will keep the KB up-to-date automatically, no, not at the moment, but as of the date of submission, all databases are up-to-date that were used to construct the KB. For a while, @MenggeZhao and I will not likely update the KB because we are working on reviewer comments. After paper submission we may construct a new KB in the future or create a job to keep the KB up-to-date.

If you have suggestions for how to do this that would be appropriate for a Github issue, or a pull request you would like to create that could perhaps implement this in some way, it would be much appreciated.

[PabloBotas]

Thanks for the quick response! Yes, that is what I was thinking.

I think simply running Phenolyzer on each HPO term individually and collecting seed genes to feed into the KB. Then what we'd need to do is to download updated versions of clinvar, etc... Initially manually triggered I guess.

I sadly cannot dedicate much time for this :(

[PabloBotas]

Hi Jim! We might have time to help in designing a procedure to update the KB. Has this been taken care of since our last chat?

[jimhavrilla]

No it has not. What do you have in mind?

Jim Havrilla

On Tue, Jun 30, 2020, 10:27 AM Pablo Botas notifications@github.com wrote:

Hi Jim! We might have time to help in designing a procedure to update the KB. Has this been taken care of since our last chat?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/WGLab/Phen2Gene/issues/15#issuecomment-651828518, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABSDYBEFMJ3SHITXTC5Y7PTRZHY6VANCNFSM4K7ZE7YQ .

[PabloBotas]

I wanted to first update the DB in Phenolyzer and then from there create the KB for Phen2Gene. Ideally this should be done automatically every now and then, but at this point the MVP would be a pipeline for manual update. I'll send you an email so we can discuss the details

[jimhavrilla]

If you have a way to do this so we can run Travis CI or something like that, that would be great. Let me know.

Jim Havrilla

On Wed, Jul 1, 2020, 9:29 AM Pablo Botas notifications@github.com wrote:

I wanted to first update the DB in Phenolyzer and then from there create the KB for Phen2Gene. Ideally this should be done automatically every now and then, but at this point the MVP would be a pipeline for manual update. I'll send you an email so we can discuss the details

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/WGLab/Phen2Gene/issues/15#issuecomment-652419116, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABSDYBFIS7PIMMP4TVKZIJ3RZM25HANCNFSM4K7ZE7YQ .

[s-batalov]

It would be very helpful to formalize a workflow for KB update. We would be happy to run it quarterly (e.g. fetch the latest association lists from public source, run local phenolyzer for each term, some more precomputes, whatever is needed). If you could please simply formalize the process? Thank you in advance.

[jimhavrilla]

If we are to formalize the process, the trivial part is running it. CI or Cron jobs can run it monthly. Some links change. Could be quite difficult to formalize. We can try and bring in @MenggeZhao on this as he was the one primarily involved in building the knowledgebase.

Jim Havrilla

On Sun, Jul 12, 2020, 3:53 AM s-batalov notifications@github.com wrote:

It would be very helpful to formalize a workflow for KB update. We would be happy to run it quarterly (e.g. fetch the latest association lists from public source, run local phenolyzer for each term, some more precomputes, whatever is needed). If you could please simply formalize the process? Thank you in advance.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/WGLab/Phen2Gene/issues/15#issuecomment-657188096, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABSDYBBOIEPI764L47ESONLR3FTWTANCNFSM4K7ZE7YQ .

[jimhavrilla]

Or by formalize do you mean write a short paragraph or two about what the process is. I assume you meant code but could be wrong.

Jim Havrilla

On Sun, Jul 12, 2020, 3:55 AM Jim Havrilla semjaavria@gmail.com wrote:

If we are to formalize the process, the trivial part is running it. CI or Cron jobs can run it monthly. Some links change. Could be quite difficult to formalize. We can try and bring in @MenggeZhao on this as he was the one primarily involved in building the knowledgebase.

Jim Havrilla

On Sun, Jul 12, 2020, 3:53 AM s-batalov notifications@github.com wrote:

It would be very helpful to formalize a workflow for KB update. We would be happy to run it quarterly (e.g. fetch the latest association lists from public source, run local phenolyzer for each term, some more precomputes, whatever is needed). If you could please simply formalize the process? Thank you in advance.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/WGLab/Phen2Gene/issues/15#issuecomment-657188096, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABSDYBBOIEPI764L47ESONLR3FTWTANCNFSM4K7ZE7YQ .

[s-batalov]

Either of the two would be greatly appreciated.

Your team is in the great position as they already built it at least once and if we could take a skeleton/vignette that shows how it is typically built (with understanding that paths to assets, indeed, might change, and their formats/structures might change) we could try to adapt. Parallelize and conquer 👍

P.S. To give a small use case - I have run Phen2Gene on a thousand of cases that have been worked on at our hospital, and (just one example) the team that worked on a particular case reported back that KARS1 is ranked near the bottom (13000-ish) while it was associated with the phenotype. After checking the latest HPO tables this indeed appears to be true, but using last year's tables, it is not. So the new associations get reported every month and if we were able to keep up with updates - we would love to!

Thank you! S.B.

[jimhavrilla]

We're gonna flesh this out with Pablo, our co-author and give an update in the near future. Stay tuned.

[themkdemiiir]

Hi Jim!

Could you provide a status report on this matter? Additionally, I'd like to revise the Knowledge Base. If you have any scripts available for modifying Phenolyzer outputs corresponding to each HPO term, I would be happy to use them. Appreciate your assistance.