Mutated Genes table does not calculate frequencies correctly

[schultzn]

I see this issue in the AACR GENIE study when selecting samples that have mutations in multiple different genes. I see frequencies greater than 100% sometimes.

I think this happens when selecting genes that are only covered in some samples.

[zhx828]

@ersinciftci @onursumer could one of you take a look this ticket? If you go to genie portal, select NUTM1, ZFHX4 from Mutated Genes table, then click Select Sample you will see the endpoint returns frequency: 121.57 for NUTM1.

[zhx828]

@schultzn @jjgao
We found the issue. Use NUTM1 as example, NUTM1 is showing up in DFCI-ONCOPANEL-3, UCSF-NIMV4, WAKE-CA-NGSQ3 three panels (http://www.cbioportal.org/genie/study?id=genie_public&filters={%22mutatedGenes%22:[{%22entrezGeneIds%22:[256646]}],%22studyIds%22:[%22genie_public%22]})

But actually NUTM1 is not listed under panel DFCI-ONCOPANEL-3 . The reason it's showing up in the study view is because it has fusion data which in current way of importing fusion, we duplicate the fusion partners.

Study view uses the # of samples in panel as denominator for the frequency, but getting all # of mutations including duplicated fusions.

You can see the same thing in oncoprint http://www.cbioportal.org/genie/results/oncoprint?session_id=5c410d51e4b05228701fbd2e

[schultzn]

Got it - thanks. Another reason why we need to decouple fusions from mutation…

…

On Jan 17, 2019, at 6:28 PM, Hongxin ***@***.***> wrote: @schultzn <https://github.com/schultzn> @jjgao <https://github.com/jjgao> We found the issue. Use NUTM1 as example, NUTM1 is showing up in DFCI-ONCOPANEL-3, UCSF-NIMV4, WAKE-CA-NGSQ3 three panels (http://www.cbioportal.org/genie/study?id=genie_public&filters={%22mutatedGenes%22:[{%22entrezGeneIds%22:[256646]}],%22studyIds%22:[%22genie_public%22]} <http://www.cbioportal.org/genie/study?id=genie_public&filters=%7B%22mutatedGenes%22:%5B%7B%22entrezGeneIds%22:%5B256646%5D%7D%5D,%22studyIds%22:%5B%22genie_public%22%5D%7D>) But actually NUTM1 is not listed under panel DFCI-ONCOPANEL-3 . The reason it's showing up in the study view is because it has fusion data which in current way of importing fusion, we duplicate the fusion partners. Study view uses the # of samples in panel as denominator for the frequency, but getting all # of mutations including duplicated fusions. You can see the same thing in oncoprint http://www.cbioportal.org/genie/results/oncoprint?session_id=5c410d51e4b05228701fbd2e <http://www.cbioportal.org/genie/results/oncoprint?session_id=5c410d51e4b05228701fbd2e> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#5613 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ALV34Ho_s-cmkWihS4iX6cBl95x9sXwVks5vEQcqgaJpZM4aCVN->.

[jjgao]

Separate fusions from mutation would fix the issue. It'll take some time before we separate them completely.

For the moment, maybe we can create a Fusion Genes table in study view by using all the fusions in mutations table and for Mutated Genes table, let's not pull fusion data? @schultzn @zhx828

[schultzn]

That would be good!

…

On Jan 18, 2019, at 4:19 PM, JianJiong Gao ***@***.***> wrote: Separate fusions from mutation would fix the issue. It'll take some time before we separate them completely. For the moment, maybe we can create a Fusion Genes table in study view by using all the fusions in mutations table and for Mutated Genes table, let's not pull fusion data? @schultzn <https://github.com/schultzn> @zhx828 <https://github.com/zhx828> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#5613 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ALV34EEjK2g0WVAmyNeaSaTsjA19ltpHks5vEjpFgaJpZM4aCVN->.

[jjgao]

@zhx828 @onursumer

• do not return FUSIONS in the mutated genes API
• add an API to return FUSION gene count
• add Fusion Genes table in study view
  • in this table, not sure if we can have the Freq column since we don't have proper gene panels for fusion.

[jjgao]

@khzhu do you think you'll have the bandwidth to handle this one?

[khzhu]

yes, will start to work on it next week.

[n1zea144]

Hi @khzhu Any progress on this? It was reported by another GENIE user recently.

[khzhu]

Hi @n1zea144 , was told to hold till JJ came back. Hongxin and myself had some discussion on how to implement web APIs to retrieve the Fusion data from the database. Fusion currently is stored in the mutation/mation_event table, so it does not has it own business models, web services would still have to use those mutation repositories.
I will ask @jjgao if we could schedule a meeting to discuss the issue.

[zhx828]

• Make sure https://www.cbioportal.org/api/swagger-ui.html#/Study_View/fetchMutatedGenesUsingPOST will only return mutations (it's under internal set of endpoints)
• Add an new endpoint /fusions/fetch under internal study view returning the fusion data
• The sql query should support the filtering, by default, returning everything(this is important for other pages)
• Follow the data stream of endpoint /mutated-genes/fetch to find out how the mutation frequency is calculated and do the same thing for fusion. There may be some discussion need to have about how the fusion frequency should be calculated, but you could leave it as TODO until the later phase of implementation

[zhx828]

@khzhu let me know if you have more questions

[khzhu]

thanks, @zhx828! will take a look and get back to you.

[jjgao]

A tricky one is gene panel. We don't have proper gene panel for fusions because the gene panel for mutations are not the same as fusions. And also fusion partners are everywhere.

One option is ignore gene panels. The problem will be that the frequencies will be lower than the actual number.

Moveover, in GENIE, only a subset has fusion data. Maybe we can create a fusion case list and use that as the denominator? @n1zea144

@schultzn thoughts?

[khzhu]

thanks, @jjgao ! agree. it might be a good idea to set gene panel aside and see if the new fusion api will resolve the issue Niki reported first.

[n1zea144]

It would be relatively easy for us to create a fusion case list for IMPACT samples. Seems like a good way to restrict the mutation frequency to a pool of samples where fusion calls were attempted to be made.

[khzhu]

Hi @jjgao @zhx828 , here is my plan:

1. backend:

• keep existing myBatis query getSampleCountInMultipleMolecularProfiles as it is, since it has a where clause used by many other myBatis queries
• add a new myBatis query getSampleCountInMultipleMolecularProfilesForMutationOrFusion to pull Mutations Or Fusions out from the mutation table.
• add a new web service getSampleCountInMultipleMolecularProfilesForFusion
• add a new API endpoint /fusion-genes/fetch

2. frontend:

• add a new Fusion table without Frequency column in the study view

What do you think? Thanks!

[jjgao]

@n1zea144 it would be great if you can priotize that. The case list ID can probably be [study_id]_fusion.

@khzhu if the fusion case list exists, please use it (its overlapping samples with selected samples) as the denominator; otherwise, use all selected samples.

[khzhu]

@JJ, will work on fusion case list with @n1zea144 . thanks!

[jjgao]

@khzhu I think you can assume _fusion case list may exist. Once @n1zea144 creates the case list, it'll work.

[khzhu]

@jjgao , okay, thanks!

[khzhu]

@jjgao @zhx828 , quick updates on this. I got backend part done, now working on the frontend, adding the new fusion table.
Have a quick question: since we filter out fusions from muteded_genes, those fusions will not show up in the oncoprint is that okay?

[zhx828]

I think the decision is leave other pages unaffected, only filter out fusions out of mutated genes table in study view.

…

On Fri, May 10, 2019 at 5:21 PM Kelsey Zhu ***@***.***> wrote: @jjgao <https://github.com/jjgao> @zhx828 <https://github.com/zhx828> , quick updates on this. I got backend part done, now working on the frontend, adding the new fusion table. Have a quick question: since we filter out fusions from muteded_genes, those fusions will not show up in the oncoprint is that okay? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#5613 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABJGQF6TGURUIG2FA3T3A53PUXRMDANCNFSM4GQJKN7A> . ===================================================================== Please note that this e-mail and any files transmitted from Memorial Sloan Kettering Cancer Center may be privileged, confidential, and protected from disclosure under applicable law. If the reader of this message is not the intended recipient, or an employee or agent responsible for delivering this message to the intended recipient, you are hereby notified that any reading, dissemination, distribution, copying, or other use of this communication or any of its attachments is strictly prohibited. If you have received this communication in error, please notify the sender immediately by replying to this message and deleting this message, any attachments, and all copies and backups from your computer.

[khzhu]

seems that query is used by oncoprint as well.

[khzhu]

@jjgao @zhx828 , I've completed all tasks as planned. Now, the Mutated Gene table excludes all fusions (please see attached, the one on the left is the one excluding all fusions. I used breast_msk_2018 dataset for testing), while new Fusion Genes table lists all fusions (please see attached). No other pages such as oncoprint are affected.
I had problems to get the genie dataset imported to my local test database (downloaded data from genie site, but somehow the clinical data was not the up-to-date one). So, if the fusions were the root cause then this fixing should resolve the issue @schultzn reported.