Multi-gene searches

[logmosier]

Right now if the user enters multiple genes in the search the landing box does not appear and the user does not have the ability to access the interactions app. A better solution to this situation should be implemented.
I think the landing box could look something like this

When the user clicks on more they will be shown the content that would come up from a search of only that gene.

The interactions app could either show the paths between the genes searched or could show the nodes from both of the neighbourhoods interweaved.

Paths between:
pros- would likely be a smaller network,
cons- differs from the behaviour of single gene searches which could be confusing

Multiple neighbourhoods:
pros-in line with current behaviour, not likely to have any unexpected/confusing behaviour for the user
cons-the network could be extremely large

Also with in the Interactions app I think there should be a way for the user to add new queried genes to the view. This could be added to the filter menu or could be included in its own menu accesed through the toolbar.

[jvwong]

Nice! I can see this working.
One reservation with multiple neighbourhoods is the 'interweave' part - that sounds like 'merging' and could get ugly. I think for pathsbetween you can set a limit:

limit= [Optional] graph query search distance limit (default = 1).

Your add gene idea is interesting, and starts to sound a lot like GeneMania was intended to do.

Also I'm gonna leave this 'big picture' idea, that kinda looks at all of this, including @wxli0 's enrichment services as all different aspects of search...

[logmosier]

Waiting on PR #535

[logmosier]

The one problem I can think of is the pathsbetween gives some results that may confuse the user. For instance if you search for ATM and MDM2 which have networks like


you get a network like

from pathsBetween. This is without setting limit so it may be better by adjusting limit but as a user I think this would be confusing why the view with both of them is much smaller.

[jvwong]

Let's assume that the user types in two genes in hopes of answering the question: "Are there known instances where these genes (directly) interact?". In that case, I believe the pathsbetween result is spot on, with one small exception...

I looked at the SIF responses in Cytoscape for either ATM or MDM2 (neighborhood) and both (pathsbetween). I stripped out everything but the two proteins.

A) ATM

B) MDM2

C) ATM and MDM2

So the dual gene search drops the interacts-with (for some reason? @IgorRodchenkov) but, looks great to me. In fact, you could probably skip the trimming step here if you want.

[IgorRodchenkov]

As far as I recall (@ozgunbabur can correct me if I am wrong), PC PATHSBETWEEN, PATHSFROMTO graph query algorithms does not take into account PPIs, i.e., MolecularInteraction interactions (thus PSI-MI data sources).

…

On Wed, Mar 21, 2018 at 5:05 PM, Jeffrey ***@***.***> wrote: Let's assume that the user types in two genes in hopes of answering the question: "Are there known instances where these genes (directly) interact?". In that case, I believe the pathsbetween result is spot on, with one small exception... I looked at the SIF responses in Cytoscape for either ATM or MDM2 (neighborhood) and both (pathsbetween). I stripped out everything but the two proteins. A) ATM <http://www.pathwaycommons.org/pc2/graph?source=http://identifiers.org/uniprot/Q13315&kind=neighborhood&format=txt> [image: image] <https://user-images.githubusercontent.com/4706307/37737167-629a3984-2d29-11e8-82e5-643678917b5b.png> B) MDM2 <http://www.pathwaycommons.org/pc2/graph?source=http://identifiers.org/uniprot/Q00987&kind=neighborhood&format=txt> [image: image] <https://user-images.githubusercontent.com/4706307/37737244-8ee2f742-2d29-11e8-99b4-44ecaeb96f96.png> C) ATM and MDM2 <http://www.pathwaycommons.org/pc2/graph?source=http://identifiers.org/uniprot/Q13315&source=http://identifiers.org/uniprot/Q00987&kind=PATHSBETWEEN&format=txt> [image: image] <https://user-images.githubusercontent.com/4706307/37737285-a9ac02a8-2d29-11e8-85a1-8430752d7aa9.png> So the dual gene search drops the interacts-with (for some reason? @IgorRodchenkov <https://github.com/igorrodchenkov>) but, looks great to me. In fact, you could probably skip the trimming step here if you want. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#520 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA8fweXdnQuPLo75UY09LQSRfBUZkmP-ks5tgsCVgaJpZM4Ssp70> .

[IgorRodchenkov]

Off-topic:
also mind issue #438 that depends on PathwayCommons/cpath2#284 and BioPAX/Paxtools#34 - both are fixed but won't have effect until the next PC release (PC10).
Hopefully, neighborhood query results won't be so large.

[ozgunbabur]

Yes, those graph queries are very process-centric. Their definition of "distance" is based on the number of Conversions one has to hop over to link two molecules. On Wed, Mar 21, 2018 at 2:12 PM, Igor Rodchenkov <[email protected]> wrote:

…

As far as I recall ***@***.*** can correct me if I am wrong), PC PATHSBETWEEN, PATHSFROMTO graph query algorithms does not take into account PPIs, i.e., MolecularInteraction interactions (thus PSI-MI data sources). On Wed, Mar 21, 2018 at 5:05 PM, Jeffrey ***@***.***> wrote: > Let's assume that the user types in two genes in hopes of answering the > question: "Are there known instances where these genes (directly) > interact?". In that case, I believe the pathsbetween result is spot on, > with one small exception... > > I looked at the SIF responses in Cytoscape for either ATM or MDM2 > (neighborhood) and both (pathsbetween). I stripped out everything but the > two proteins. > > A) ATM > <http://www.pathwaycommons.org/pc2/graph?source=http:// identifiers.org/uniprot/Q13315&kind=neighborhood&format=txt> > [image: image] > <https://user-images.githubusercontent.com/4706307/ 37737167-629a3984-2d29-11e8-82e5-643678917b5b.png> > > B) MDM2 > <http://www.pathwaycommons.org/pc2/graph?source=http:// identifiers.org/uniprot/Q00987&kind=neighborhood&format=txt> > [image: image] > <https://user-images.githubusercontent.com/4706307/ 37737244-8ee2f742-2d29-11e8-99b4-44ecaeb96f96.png> > > C) ATM and MDM2 > <http://www.pathwaycommons.org/pc2/graph?source=http:// identifiers.org/uniprot/Q13315&source=http://identifiers.org/uniprot/ Q00987&kind=PATHSBETWEEN&format=txt> > [image: image] > <https://user-images.githubusercontent.com/4706307/ 37737285-a9ac02a8-2d29-11e8-85a1-8430752d7aa9.png> > > So the dual gene search drops the interacts-with (for some reason? > @IgorRodchenkov <https://github.com/igorrodchenkov>) but, looks great to > me. In fact, you could probably skip the trimming step here if you want. > > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub > <https://github.com/PathwayCommons/app-ui/issues/ 520#issuecomment-375095338>, > or mute the thread > <https://github.com/notifications/unsubscribe-auth/ AA8fweXdnQuPLo75UY09LQSRfBUZkmP-ks5tgsCVgaJpZM4Ssp70> > . > — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#520 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABUCCyRyAXVnbAgugOoX2Fhaaskg_0o2ks5tgsI0gaJpZM4Ssp70> .