bcm-uga / hadaca3

0 stars 1 forks source link

Unable to identify issue #28

Closed lionel-spinelli closed 1 month ago

lionel-spinelli commented 1 month ago

Hello, I have added a simple code to perform feature selection in the Program function. To run this code, I need to add the MatrxStats package. The code runs in my docker environment but fails in the submission on codabench. Scoring logs reports :

Error in gzfile(file, "rb") : cannot open the connection
Calls: readRDS -> gzfile
In addition: Warning message:
In gzfile(file, "rb") :
  cannot open compressed file '/app/input/res/prediction.rds', probable reason 'No such file or directory'
Execution halted

In the reports I have no information to understand what's wrong. I have added some text traces to follow the process but those traces are not reported in the report page neither in Logs-> Stdout tabs or elsewhere.

The last submission ID that should work is 85381. Could you please tell me how to debug this issue ? Thanks

Here is the piece of code I have added:

 if ( !( "matrixStats" %in% installed.packages() ) ) {
    install.packages(pkgs = "matrixStats")
  } 

  # Select the top feature for RNA
  require( matrixStats)

  cat("\nComputing Variance on RNA data")
    rna_variance = matrixStats::rowVars( mix_rna_orig)

  cat("\nComputing Top genes on RNA data")
    top_rna_gene = names( rna_variance)[ order( rna_variance, decreasing = TRUE)[1:1000]]

  mix_rna = mix_rna_orig[ top_rna_gene,]
  ref_rna = ref_rna_orig[ top_rna_gene,]

  # Select the top feature for methylation
  cat("\nComputing Variance on methylation data")
  met_variance = rowVars( mix_met_orig)

  cat("\nComputing Top features on methylation data")
  top_met_feature = names( met_variance)[ order( met_variance, decreasing = TRUE)[1:1000]]

  mix_met = mix_met_orig[ top_met_feature,]
  ref_met = ref_met_orig[ top_met_feature,] 
nicomy commented 1 month ago

Hello Lionel,

Can you provide the URL of this challenge ? :)

You read the logs of the scoring program which indicate that the file prediction.rds is not found, which means that the execution of your code failed.

Can you read the logs of the prediction program ?

To access it, you can, click on your submission listed in the table under my submissions, and then go into logs and read the four different logs : "stdout, stderr, Ingestion stdout, Ingestion stderr" .

lionel-spinelli commented 1 month ago

Hello Nicolas,

thanks for your response. The problem is that there is no message in the prediction section (apart the ingestion that recapitulate some system information). The only messages are in the scoring log.

Do you want me to join the zip files provided by the "Download" but for prediction and scoring ?

Best

——————————————————————————— [1727425308686]


De : Nicolas Homberg @.***> Envoyé : mardi 1 octobre 2024 14:28:28 À : bcm-uga/hadaca3 Cc : Lionel Spinelli; Author Objet : Re: [bcm-uga/hadaca3] Unable to identify issue (Issue #28)

Ce mail provient de l'extérieur, restons vigilants

Hello Lionel,

Can you provide the URL of this challenge ? :)

You read the logs of the scoring program which indicate that the file prediction.rds is not found, which means that the execution of your code failed.

Can you read the logs of the prediction program ?

To access it, you can, click on your submission listed in the table under my submissions, and then go into logs and read the four different logs : "stdout, stderr, Ingestion stdout, Ingestion stderr" .

— Reply to this email directly, view it on GitHubhttps://github.com/bcm-uga/hadaca3/issues/28#issuecomment-2385646280, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ALFWT62PNWHXJ6VPEQWMPXTZZKIOZAVCNFSM6AAAAABPFNZSBOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGOBVGY2DMMRYGA. You are receiving this because you authored the thread.Message ID: @.***>

nicomy commented 1 month ago

Can you share more details please?

* URL of the competition (we have different versions of the hadaca3 ^) 
* Your username
* The submission ID ? 

Thank you :)

lionel-spinelli commented 1 month ago

Hello,

thanks for your message. Here is the URL of the competition :

https://www.codabench.org/competitions/3032/?secret_key=28767dd4-3051-4d5c-8d25-13d0922481fb#/participate-tab

My username is "spinelli"

Th submission ID is 85381

Thanks you

Best

——————————————————————————— [1727425308686]


De : Nicolas Homberg @.***> Envoyé : jeudi 3 octobre 2024 08:21:16 À : bcm-uga/hadaca3 Cc : Lionel Spinelli; Author Objet : Re: [bcm-uga/hadaca3] Unable to identify issue (Issue #28)

Ce mail provient de l'extérieur, restons vigilants

Can you share more details please?

Thank you :)

— Reply to this email directly, view it on GitHubhttps://github.com/bcm-uga/hadaca3/issues/28#issuecomment-2390614523, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ALFWT63EYM4KTHLCAU6RUODZZTO5ZAVCNFSM6AAAAABPFNZSBOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGOJQGYYTINJSGM. You are receiving this because you authored the thread.Message ID: @.***>

nicomy commented 1 month ago

Hello Lionel, I have news :)

I studied your submissions and I can read in the ingestion step :

[1] "/app/ingested_program/program.R"
[1] "no program to execute, go straight to scoring step"

Your submission file contains : submission_script.R and should contain a file named program.R.

In order to create this program.R, the recommended way is to execute on your computer submission_script.R with the command Rscript submision_script.R.

This script will create inside a folder submissions two zip folders that are ready to be submitted on Codabench :

The second possibility (less recommended!) is to manually create your program.R inside a zip file.

Tell me if you have further problems !

Also there are more recent challenges here : Introductory Smoothie challenge Phases 2-3 of hadaca3 :

There are some other useful tutorials on these challenges ! I would recommend starting with the Smoothie challenge, because it deals with a much smaller data set :)

Have a good day, and feel free to reopen this issue if you need further assistance. Nico

lionel-spinelli commented 1 month ago

Hello,

thanks for the analysis. In fact Magalia lso told me to use the newest version of the codabench. So I have used the version : https://www.codabench.org/competitions/3742/?secret_key=d7080226-30d4-48f1-934d-599aa0b6ff72#/pages-tab

And following your instruction, it works when I submit the program.zip file generated by the submission_script.R. However, when I submit the results.zip, I have a failed result (see submission 87344 and 88850).

Thanks again and sorry to do not have correctly the tutorials.

Best

——————————————————————————— [1727425308686]


De : Nicolas Homberg @.***> Envoyé : mardi 8 octobre 2024 11:31:52 À : bcm-uga/hadaca3 Cc : Lionel Spinelli; Author Objet : Re: [bcm-uga/hadaca3] Unable to identify issue (Issue #28)

Ce mail provient de l'extérieur, restons vigilants

Hello Lionel, I have news :)

I studied your submissions and I can read in the ingestion step :

[1] "/app/ingested_program/program.R" [1] "no program to execute, go straight to scoring step"

Your submission file contains : submission_script.R and should contain a file named program.R.

In order to create this program.R, the recommended way is to execute on your computer submission_script.R with the command Rscript submision_script.R.

This script will create inside a folder submissions two zip folders that are ready to be submitted on Codabench :

The second possibility (less recommended!) is to manually create your program.R inside a zip file.

Tell me if you have further problems !

Also there are more recent challenges here : Introductory Smoothie challenge https://www.codabench.org/competitions/3743/?secret_key=6f226b2e-78c0-42e9-bb80-a8bdc47af475 Phases 2-3 of hadaca3 : https://www.codabench.org/competitions/3742/?secret_key=d7080226-30d4-48f1-934d-599aa0b6ff72

There are some other useful tutorials on these challenges ! I would recommend starting with the Smoothie challenge, because it deals with a much smaller data set :)

Have a good day, and feel free to reopen this issue if you need further assistance. Nico

— Reply to this email directly, view it on GitHubhttps://github.com/bcm-uga/hadaca3/issues/28#issuecomment-2399343168, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ALFWT62LYTQY6O6TFA2HLC3Z2ORARAVCNFSM6AAAAABPFNZSBOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGOJZGM2DGMJWHA. You are receiving this because you authored the thread.Message ID: @.***>