search

callahantiff / PheKnowLator

PheKnowLator: Heterogeneous Biomedical Knowledge Graphs and Benchmarks Constructed Under Alternative Semantic Models
https://github.com/callahantiff/PheKnowLator/wiki
Apache License 2.0
159 stars 29 forks source link

✅ - Need pkt Build Framework Overview Figure #79

Closed callahantiff closed 3 years ago

callahantiff commented 3 years ago

TASK

Need to create a figure that provides an overview of the pkt knowledge graph construction framework

TODO

callahantiff commented 3 years ago

Overview / Figure 1 to describe PheKnowLator is now complete! @LEHunter @bill-baumgartner @mgkahn, what do you think (click to enlarge image)?

Figure1_Overview

mgkahn commented 3 years ago

Four comments by Kahn PhenKnowLator_figure1_Kahn

callahantiff commented 3 years ago

Four comments by Kahn PhenKnowLator_figure1_Kahn

Thanks for the detailed feedback @mgkahn , the comment bubbles were super clear! Responding to your comments left to right:

  1. The body image under the "Annotation/Normalization" section is supposed to demonstrate that the process includes data that we have annotated. In this case the figure is meant to demonstrate the "hTERT" cell type for mammary gland cells and how each of the "parts" involved has been annotated to the UBERON anatomy, cell (CL), and cell line (CLO) ontologies. It's meant to show experimental data (in this case) is converted into something that can be integrated with ontologies. Do I need to add something else to make that more clear?
  2. You are right, that's confusing. I will remove that arrow. There is already an arrow coming from the "Knowledge Graph Construction" square and that is the correct one.
  3. I will remove obo there. That's a good point. Or, can I update the example to use other namespaces in addition to obo. Would that be more clear?
  4. The removal of triples piece is correct, but technically it's hyperparameter in that you can choose not to remove triples, but if you do choose to remove triples, you can choose the method for how they are removed. Do I need to add something else to make that more clear?

The updated figure for comments 2 and 3 is included below:
Figure1_Overview

mgkahn commented 3 years ago

For semantic abstraction: If there are alternatives (maybe implied by two downward arrows), then maybe show two different result graph structures at the end of each downward arrow. Confusing why you have two arrows leading to the single result graph.

For Annotation/Normalization: No way I would have gotten what you described from the figure but perhaps somebody who knows the ontologies would understand. But this figure it supposed to "speak" to the unknowledgeable so give some thought to another way of capturing what you wrote. This figure doesn't work for the uneducated.

And its @mgkahn , not @LEHunter who is the uneducated one here.

callahantiff commented 3 years ago

For semantic abstraction: If there are alternatives (maybe implied by two downward arrows), then maybe show two different result graph structures at the end of each downward arrow. Confusing why you have two arrows leading to the single result graph.

For Annotation/Normalization: No way I would have gotten what you described from the figure but perhaps somebody who knows the ontologies would understand. But this figure it supposed to "speak" to the unknowledgeable so give some thought to another way of capturing what you wrote. This figure doesn't work for the uneducated.

And its @mgkahn , not @LEHunter who is the uneducated one here.

Sorry about the confusion in my last post. I accidentally tagged Larry when I meant to tag you. OK, the next pass on the figure is attached below. I have updated the semantic abstraction section to address the fact the double arrow problem. I can see how that was confusing. I will have to think about the annotation/normalization part. I think this is an important contribution that is provided so I don't want to remove that section, but I hear what you are saying about it not being sufficient for someone without domain knowledge. I can follow-up with Larry, he might have some other ideas.

Figure1_Overview

bill-baumgartner commented 3 years ago

fyi - My comments are based on the previous version of the figure as I didn't see this new one in time. pkt-build

callahantiff commented 3 years ago

fyi - My comments are based on the previous version of the figure as I didn't see this new one in time. pkt-build

Thanks @bill-baumgartner for the additional feedback! I will respond to the comments as they are organized by each square on the figure. Hope you can read it OK 😄

  1. Build Edge Lists: The use of rectangles and color are really meant to be a simplified way of showing that in order to get to the edge list on the right some of the initial identifiers have to be updated. I could try to re-work this to better demonstrate that?
  2. Merge Ontologies: I was trying to make it very obvious how the figures above the line would fit together to make the bottom image. Do you recommend something else? I could add some circles to highlight the specific merge points in addition to the coloring if that would be more clear?
  3. Select Build Hyperparameters:
    • It's just meant to show that there are two ways to model knowledge and they instance-based and subclass-based. Is it confusing as I have done it? It does actually correctly map onto the simplest version of our patterns 😉
    • I will try and squeeze the strategy names in there. That's a good idea!
    • Fixed. Going back to just one output, but with a single arrow
  4. Data Preparation:
    • Awesome, thank you! Yes, that's exactly what the dotted line means! Is it too much here?
    • Excellent suggestion, that's a great idea and I bet might help make it more clear what we are doing there.
    • OK, that's a good point. Something like "ID" or "Entity ID"? Let me know what you think about the change I made.
  5. Knowledge Graphs: I was trying to use the blue line to indicate all 3 nodes and the red line to indicate each entity (s, p, o) in the triple. I could separate this into 2 figures if that would be more clear?
  6. Public Endpoints: Cloud file storage it is!


OK updated figure below to reflect your suggestions. better? Figure1_Overview

callahantiff commented 3 years ago

Got some additional great feedback from @mgkahn and @LEHunter. Will be working those changes today.

callahantiff commented 3 years ago

@mgkahn and @LEHunter -- please updated figure with your requested changes. Note that there is still some light wordsmithing needed, specifically around how to refer to the output that's generated by the ecosystem/system/platform. Our goal is to make sure we convey that there are multiple kinds of/multimodal output generated. Otherwise, all other changes have been made. Thanks again for your feedback, I think it's really coming along nicely.

Figure1_Overview

mgkahn commented 3 years ago

Without your previous explanation, I still do not "get" Concept Annotation from the diagram. Maybe we just move on. The rest of the figure is good. Like how you redid Semantic Abstraction.

callahantiff commented 3 years ago

Without your previous explanation, I still do not "get" Concept Annotation from the diagram. Maybe we just move on. The rest of the figure is good. Like how you redid Semantic Abstraction.

Thank you for taking another look at it. I'm happy to keep thinking about ways to improve the Concept Annotation square. I do think your feedback this week has helped to improve it. I don't think this resolve your concerns, but I did slightly update that part in the version below. Will stop fiddling now see what Larry and Bill think as well.

Figure1_Overview

mgkahn commented 3 years ago

Time to move onto other higher priority tasks........

callahantiff commented 3 years ago

@LEHunter and @bill-baumgartner I will close this issue, please re-open if there are additional changes you would like made to the figure.

LEHunter commented 3 years ago

One last comment. I like the rearrangement, but wish there were a way to continue the arrows running left to right across the top. I didn’t even see the arrow pointing from KG construction to generated output at first. Maybe move the endpoints and benchmarks below generated output, and run one arrow down the right edge?

On Apr 17, 2021, at 8:49 AM, Tiffany J. Callahan @.***> wrote:

Without your previous explanation, I still do not "get" Concept Annotation from the diagram. Maybe we just move on. The rest of the figure is good. Like how you redid Semantic Abstraction.

Thank you for taking another look at it. I'm happy to keep thinking about ways to improve the Concept Annotation square. I do think your feedback this week has helped to improve it. I don't think this resolve your concerns, but I did slightly update that part in the version below. Will stop fiddling now see what Larry and Bill think as well.

https://user-images.githubusercontent.com/8030363/115123640-3c83da80-9f7b-11eb-8e40-294467db4bb6.png — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/callahantiff/PheKnowLator/issues/79#issuecomment-821869569, or unsubscribe https://github.com/notifications/unsubscribe-auth/AACWZKL673V4BPEJZTGE72TTJHJ2VANCNFSM4VKGQNUQ.

callahantiff commented 3 years ago

One last comment. I like the rearrangement, but wish there were a way to continue the arrows running left to right across the top. I didn’t even see the arrow pointing from KG construction to generated output at first. Maybe move the endpoints and benchmarks below generated output, and run one arrow down the right edge? … On Apr 17, 2021, at 8:49 AM, Tiffany J. Callahan @.***> wrote: Without your previous explanation, I still do not "get" Concept Annotation from the diagram. Maybe we just move on. The rest of the figure is good. Like how you redid Semantic Abstraction. Thank you for taking another look at it. I'm happy to keep thinking about ways to improve the Concept Annotation square. I do think your feedback this week has helped to improve it. I don't think this resolve your concerns, but I did slightly update that part in the version below. Will stop fiddling now see what Larry and Bill think as well. https://user-images.githubusercontent.com/8030363/115123640-3c83da80-9f7b-11eb-8e40-294467db4bb6.png — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#79 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AACWZKL673V4BPEJZTGE72TTJHJ2VANCNFSM4VKGQNUQ.

OK, thanks @LEHunter. I have updated the figure. I think this addresses your request, without requiring significance overall. Can we call this the one?

Figure1_Overview

LEHunter commented 3 years ago

Works for me!

On Apr 20, 2021, at 9:21 AM, Tiffany J. Callahan @.***> wrote:

One last comment. I like the rearrangement, but wish there were a way to continue the arrows running left to right across the top. I didn’t even see the arrow pointing from KG construction to generated output at first. Maybe move the endpoints and benchmarks below generated output, and run one arrow down the right edge? … <x-msg://70/#> On Apr 17, 2021, at 8:49 AM, Tiffany J. Callahan @.***> wrote: Without your previous explanation, I still do not "get" Concept Annotation from the diagram. Maybe we just move on. The rest of the figure is good. Like how you redid Semantic Abstraction. Thank you for taking another look at it. I'm happy to keep thinking about ways to improve the Concept Annotation square. I do think your feedback this week has helped to improve it. I don't think this resolve your concerns, but I did slightly update that part in the version below. Will stop fiddling now see what Larry and Bill think as well. https://user-images.githubusercontent.com/8030363/115123640-3c83da80-9f7b-11eb-8e40-294467db4bb6.png https://user-images.githubusercontent.com/8030363/115123640-3c83da80-9f7b-11eb-8e40-294467db4bb6.png — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#79 (comment) https://github.com/callahantiff/PheKnowLator/issues/79#issuecomment-821869569>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AACWZKL673V4BPEJZTGE72TTJHJ2VANCNFSM4VKGQNUQ https://github.com/notifications/unsubscribe-auth/AACWZKL673V4BPEJZTGE72TTJHJ2VANCNFSM4VKGQNUQ.

OK, thanks @LEHunter https://github.com/LEHunter. I have updated the figure. I think this addresses your request, without requiring significance overall. Can we call this the one?

https://user-images.githubusercontent.com/8030363/115451770-d8f7e800-a1da-11eb-90ac-78b3fda5dd65.png — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/callahantiff/PheKnowLator/issues/79#issuecomment-823539137, or unsubscribe https://github.com/notifications/unsubscribe-auth/AACWZKPSGSJQ6UNYAOJXK33TJXH33ANCNFSM4VKGQNUQ.

mgkahn commented 3 years ago

I also like this FINAL (!!) redesign. I also noticed the redesign of Concept Annotation. Let's go with this version.

callahantiff commented 3 years ago

I also like this FINAL (!!) redesign. I also noticed the redesign of Concept Annotation. Let's go with this version.

Sounds great. I'm officially closing this. Notice the super official checkmark emoji I added to the issue title to convey this 😄 .

callahantiff commented 3 years ago

Wait, wait, @LEHunter dare I say that this figure will satisfice?!

LEHunter commented 3 years ago

indeed. Need a new check box?

On Apr 21, 2021, at 11:32 AM, Tiffany J. Callahan @.***> wrote:

Wait, wait, @LEHunter https://github.com/LEHunter dare I say that this figure will satisfice https://www.nobelprize.org/uploads/2018/06/simon-lecture.pdf?!

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/callahantiff/PheKnowLator/issues/79#issuecomment-824371048, or unsubscribe https://github.com/notifications/unsubscribe-auth/AACWZKI4FPZMZWHGZDIQOMTTJ476TANCNFSM4VKGQNUQ.

callahantiff commented 3 years ago

indeed. Need a new check box? … On Apr 21, 2021, at 11:32 AM, Tiffany J. Callahan @.***> wrote: Wait, wait, @LEHunter https://github.com/LEHunter dare I say that this figure will satisfice https://www.nobelprize.org/uploads/2018/06/simon-lecture.pdf?! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#79 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AACWZKI4FPZMZWHGZDIQOMTTJ476TANCNFSM4VKGQNUQ.

AWESOME. Better than a checkmark we need a Herbert Simon Emoji!