nextflow-io/nextflow

Nextflow community chat moved to Slack! https://www.nextflow.io/blog/2022/nextflow-is-moving-to-slack.html

Paolo Di Tommaso

@pditommaso

it's possible for the nextflow runtime development, but I guess you want for nextflow scripts

mmatthews06

@mmatthews06

No, runtime development. I'm mucking around Nextflow internals for the time being.

Paolo Di Tommaso

@pditommaso

then it's straightforward, use ./launch.sh -remote-debug run .. etc

mmatthews06

@mmatthews06

Ah, alright, I'll try that. Thanks!

Paolo Di Tommaso

@pditommaso

Workflow components in the pipeline :sunglasses:

https://twitter.com/PaoloDiTommaso/status/1161211908554469376

Stephen Kelly

@stevekm

@banjosnapper put your scripts in a directory called bin adjacent to the main nextflow script; example here: https://github.com/stevekm/nextflow-demos/tree/1238d0c444f388cb1ee79c351a57610e03e4bbb6/R-Python

as long as the scripts are executable then you can just invoke them directly from within your task

Guillaume Theaud

@GuillaumeTh

@pditommaso I saw that now in the .command.run the values for the environment variable are written like MY_VAR=\"0\" whereas in the 19.04 the variable was MY_VAR="0". This modification is normal ?

Stephen Kelly

@stevekm

@banjosnapper instead of calling your script with perl myscript.pl you should instead put a shebang at the very first line that invokes the perl interpreter, see this Python script that includes one: https://github.com/stevekm/nextflow-demos/blob/1238d0c444f388cb1ee79c351a57610e03e4bbb6/R-Python/bin/test.py

this allows you to invoke the script as simply myscript.pl

Paolo Di Tommaso

@pditommaso

@GuillaumeTh :point_right: nextflow-io/nextflow#1146

Guillaume Theaud

@GuillaumeTh

@pditommaso thanks

@pditommaso The problem that I have in my case is that in the config file I set env {NB_PROCESSES=1} But with the fix the value of NB_PROCESSES="1" but some scripts requiered a int not a string. Do you have a idea to set a int instead of a string in the nextflow.config ?

Paolo Di Tommaso

@pditommaso

Don't think env vars distinguish between num and string

Stephen Kelly

@stevekm

sometimes when I restart a Nexflow pipeline with -resume, I get errors inside my processes such as java.lang.NullPointerException: Cannot get property 'outputDir' on null object. These come from processes that look like this:

process vcf_to_tsv {
    publishDir "${params.outputDir}/VEP/vcf_tsv", mode: 'copy'
    // stuff
}

so it seems like params is not always getting initialized correctly when the pipeline resumes.

I have the same error for dict objects that I initialize from reading in a JSON. Sometimes, the pipeline errors out when I try to access the object's keys inside a process. Problem is, all of these errors are completely random and not reproducible. Eventually if I restart the pipeline enough times with -resume, they go away. Or not, and I just start the pipeline from scratch... is this a known bug? I think we are still using 19.01

here is one such stack trace:

Dec-18 09:01:45.818 [PathVisitor-1] ERROR nextflow.Channel - Cannot get property 'outputDir' on null object
java.lang.NullPointerException: Cannot get property 'outputDir' on null object
        at org.codehaus.groovy.runtime.NullObject.getProperty(NullObject.java:60)
        at org.codehaus.groovy.runtime.InvokerHelper.getProperty(InvokerHelper.java:190)
        at org.codehaus.groovy.runtime.callsite.NullCallSite.getProperty(NullCallSite.java:46)
        at org.codehaus.groovy.runtime.callsite.GetEffectivePogoPropertySite.callGetProperty(GetEffectivePogoPropertySite.java:45)
        at _nf_script_121eedc3$_run_closure117$_closure442.doCall(_nf_script_121eedc3:2192)
        at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
        at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
        at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
        at java.lang.reflect.Method.invoke(Method.java:498)
        at org.codehaus.groovy.reflection.CachedMethod.invoke(CachedMethod.java:104)
        at groovy.lang.MetaMethod.doMethodInvoke(MetaMethod.java:326)
        at org.codehaus.groovy.runtime.metaclass.ClosureMetaClass.invokeMethod(ClosureMetaClass.java:264)
        at groovy.lang.MetaClassImpl.invokeMethod(MetaClassImpl.java:1041)
        at groovy.lang.Closure.call(Closure.java:411)
        at groovy.lang.Closure.call(Closure.java:405)
        at groovy.lang.GString.writeTo(GString.java:189)
        at groovy.lang.GString.toString(GString.java:153)
        at org.codehaus.groovy.runtime.typehandling.ShortTypeHandling.castToString(ShortTypeHandling.java:55)
        at nextflow.extension.Bolts.asType(Bolts.groovy:474)
        at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
        at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
        at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
        at java.lang.reflect.Method.invoke(Method.java:498)
        at org.codehaus.groovy.runtime.metaclass.ReflectionMetaMethod.invoke(ReflectionMetaMethod.java:54)
        at org.codehaus.groovy.runtime.metaclass.NewInstanceMetaMethod.invoke(NewInstanceMetaMethod.java:56)
        at groovy.lang.MetaMethod.doMethodInvoke(MetaMethod.java:326)
        at groovy.lang.MetaClassImpl.invokeMethod(MetaClassImpl.java:1235)
        at groovy.lang.MetaClassImpl.invokeMethod(MetaClassImpl.java:1041)
        ......

Paolo Di Tommaso

@pditommaso

it seems an app issue, outputDir is not a NF variable

Stephen Kelly

@stevekm

It is defined inside the main.nf pipeline;

params.outputDir = "output"

process vcf_to_tsv {
      publishDir "${params.outputDir}/VEP/vcf_tsv", mode: 'copy'
....
....
}

this way the user is able to override it from the CLI

nextflow run main.nf --outputDir foo/

I thought this was standard practice?

Paolo Di Tommaso

@pditommaso

weird, try the latest version, if the problem persits open an issue

Stephen Kelly

@stevekm

what was the arg to disable the ANSI logger in the latest?

Paolo Di Tommaso

@pditommaso

-ansi-log false

Stephen Kelly

@stevekm

oh sorry I meant the environment variable...

Paolo Di Tommaso

@pditommaso

NXF_ANSI_LOG=false

Stephen Kelly

@stevekm

I get a lot of these... WARN: The channel 'create' method is deprecated -- it will be removed in a future release

Ha Le

@haqle314

Yeah, disabling ANSI logger just means that nextflow will print out plain logs without color, '\r', etc. I think nextflow should print warning messages to stderr instead of stdout as it currently does. This way, people that dont want to see the warnings can do nextflow ... 2>/dev/null.

Stijn van Dongen

@micans

Is it possible not to use the time directive? In some cases I'd like to depend on whatever the (LSF) queue maximum is, perhaps in a process-specific manner. This would also avoid the problem that it is a (bsub) error to specify a time limit larger than the queue limit. And can this be done also using override, so e.g. default might be 12.h but for some process I want to turn it off.

Adrian Viehweger

@phiweger

using DSL2, how can I specify that a process uses 2 channels as input, when one channel only contains 1 item, ie channel A has "asm = [s1, s2]" and channel B has "db = [db]" and I want to perform some process with [s1, db] and [s2, db]. I tried foo(asm, db) but this starts only [s1, db] and ignores [s2, db].

Steve Frenk

@sfrenk

@phiweger Are you sure that s1 and s2 are being emitted separately rather than as a single list? eg. you may need to use flatten() or flatMap()

Has anyone had experience using Spock or JUnit (or other unit testing framework) with DSL2? I'm trying to figure out a way of getting the Specification class to have access to processes and channels.

Krittin Phornsiricharoenphant

@sinonkt

@pditommaso Just want you guys know that DSL2 is really awesome, That helps a lot Thank you so much!

@zhqu1148980644 Thank you! for pointing penv out, I never noticed this before! That’s improve readability a lot.

Krittin Phornsiricharoenphant

@sinonkt

I wonder, how can i import/include snippet of groovy code from url like github source for some of my helper function to be able to share on many project as a kind of lib.

I familiar that i found some gist that can do some kind of source from url in groovy code.

deo999

@deo999

nextflow run <pipeline name> -with-report [file name]

i want to run this but have confusion what i should give in pipeline name and file name

Alaa Badredine

@AlaaBadredine_twitter

ermm

deo999

@deo999

file name will be .nf file then what will be the pipeline name

Alaa Badredine

@AlaaBadredine_twitter

the pipeline name stands for your script.nf and the -with-report is anything u want

deo999

@deo999

what about with docker

nextflow run main.nf -with-docker -with-report result

does this commnd ok with docker

Alaa Badredine

@AlaaBadredine_twitter

-with-docker is used to run a docker image

https://www.nextflow.io/docs/latest/docker.html

deo999

@deo999

yes i know tht

Where communities thrive

People

Repo info

Activity