search

immunomind / immunarch

🧬 Immunarch: an R Package for Fast and Painless Exploration of Single-cell and Bulk T-cell/Antibody Immune Repertoires
https://immunarch.com
Apache License 2.0
297 stars 65 forks source link

Error in tSNE plot generation and automated pipeline for multiple samples , urgent help needed !! #139

Open ankitadas460 opened 3 years ago

ankitadas460 commented 3 years ago

Hello

we have an app wrapped for immune arch on DNAnexus and consistently are getting this error for many sample types and it looks like the tSNE plot doesn't like the data upstream and is throwing random errors. We are getting this error with a subset of the samples run with RTCR upstream, and would like to be able to run it as part of the applet. We have an urgent project we are working on, can we please get some assistance in working on this !!

In joined[[col_a]] joined[[col_b]] : NAs produced by integer overflow 2: In joined[[col_a]] joined[[col_b]] : NAs produced by integer overflow 3: In joined[[col_a]] * joined[[col_b]] : NAs produced by integer overflow Perplexity should be lower than K! pdf 2 Perplexity should be lower than K! pdf 2 Error in na.fail.default(X) : missing values in object Calls: vis ... Rtsne.dist -> as.matrix -> na.fail -> na.fail.default Execution halted

vadimnazarov commented 3 years ago

Hi @ankitadas460

Thank you for using Immunarch! The package is free and open source software, it's supported by our company and a part of our drug development and analytics platform. We would like to help everyone, but sadly, we are limited in resources. However, we could discuss a commercial-grade support for Immunarch if you think it would be helpful. It includes fast response rates. Please contact me at support@immunomind.io to discuss more details.

Regarding your issue. It seems that is it related to #138 , but your case have an additional issue of NAs produced by integer overflow. Do you have NAs in the Clones column in one or several of your immune repertoires?

Additionally, there is an error Perplexity should be lower than K!. When you pass the perplexity parameter, please consider it making less than the .k parameter which regulates a number of output clusters.

ankitadas460 commented 3 years ago

Ankita Das has shared a OneDrive for Business file with you. To view it, click the link below. https://miodx-my.sharepoint.com/:u:/p/adas/ERuKDABDXVFEm4OarJ0-dmsBTeA2PGRi9F7Im5SnqTwMSg [https://r1.res.office365.com/owa/prem/images/dc-generic_20.png]https://miodx-my.sharepoint.com/:u:/p/adas/ERuKDABDXVFEm4OarJ0-dmsBTeA2PGRi9F7Im5SnqTwMSg SRR7993438_1_results(1).tsvhttps://miodx-my.sharepoint.com/:u:/p/adas/ERuKDABDXVFEm4OarJ0-dmsBTeA2PGRi9F7Im5SnqTwMSg

Hello Vadim

Thanks for your response, however when I am looking at some clones files generated by RTCR I am not seeing anything labeled as NA. I have attached an examples from the ones that are part of an Immune arch run with erros , please advise what would be the next steps to dix this issue on our end.

I have also cc'd our scientific collaborator and the lead developer of the app from DNAnexus Nick Hill who can provide you with additional details if needed on the app .

Would appreciate if you could please give us some direction on how to resolve this issue as soon as possible.

Regards Ankita

From: Ankita Das @.> Sent: Saturday, April 24, 2021 7:51 AM To: immunomind/immunarch @.>; immunomind/immunarch @.> Cc: Mention @.>; Nicholas Hill @.>; Aishwarya Chander @.> Subject: Re: [immunomind/immunarch] Error in tSNE plot generation and automated pipeline for multiple samples , urgent help needed !! (#139)

Hello Vadim

Thanks for your response, however when I am looking at some clones files generated by RTCR I am not seeing anything labeled as NA. I have attached 2 examples from the ones that are part of an Immune arch run with erros , please advise what would be the next steps to dix this issue on our end.

I have also cc'd our scientific collaborator and the lead developer of the app from DNAnexus Nick Hill who can provide you with additional details if needed on the app .

Would appreciate if you could please give us some direction on how to resolve this issue as soon as possible.

Regards Ankita

From: Vadim I. Nazarov @.> Sent: Saturday, April 24, 2021 3:25 AM To: immunomind/immunarch @.> Cc: Ankita Das @.>; Mention @.> Subject: Re: [immunomind/immunarch] Error in tSNE plot generation and automated pipeline for multiple samples , urgent help needed !! (#139)

Hi @ankitadas460https://github.com/ankitadas460

Thank you for using Immunarch! The package is free and open source software, it's supported by our company and a part of our drug development and analytics platform. We would like to help everyone, but sadly, we are limited in resources. However, we could discuss a commercial-grade support for Immunarch if you think it would be helpful. It includes fast response rates. Please contact me at @.**@.> to discuss more details.

Regarding your issue. It seems that is it related to #138https://github.com/immunomind/immunarch/issues/138 , but your case have an additional issue of NAs produced by integer overflow. Do you have NAs in the Clones column in one or several of your immune repertoires?

Additionally, there is an error Perplexity should be lower than K!. When you pass the perplexity parameter, please consider it making less than the .k parameter which regulates a number of output clusters.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/immunomind/immunarch/issues/139#issuecomment-826071263, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APMP5NBEVNKOT7MQ2JKTASDTKKMCFANCNFSM43PUTNKA.

vadimnazarov commented 3 years ago

Hello Ankita!

Thank you, I'm looking into it. Please consider that by replaying to this GitHub emails you are posting messages in your ticket here https://github.com/immunomind/immunarch/issues/139 Feel free to message me on support@immunomind.io directly if the nature of the discussion is sensitive. I can also delete the messages from the ticket if you would like me to.

I explored the data and I don't see any issues with the data either. It's indeed a weird behaviour. Could you send me the output of the overlap function, as well as exact code you are trying to run with exact parameter values please? Such as 

ov = repOverlap(your_data)
ova_res = repOverlapAnalysis(ov, "tsne", .perp = 5, .k = 10)

Exact values of ov, .perp and .k will help me understand the nature of the behaviour.