Skip to content

3. Exercise 2

Move into the exercise 2 directory and execute the hello-gatk.nf script.

cd /workspace/gitpod/troubleshoot/exercise2
nextflow run hello-gatk.nf

Error message

Although there is no error message, the GATK_HAPLOTYPECALLER process is only being run once not three times as expected.

executor >  local (5)
[d2/839095] process > SAMTOOLS_INDEX (1)       [100%] 3 of 3 ✔
[1a/2dafee] process > GATK_HAPLOTYPECALLER (1) [100%] 1 of 1 ✔
[59/278fc4] process > GATK_JOINTGENOTYPING (1) [100%] 1 of 1 ✔
Solution

A common mistake by developers is having the wrong channel type.

Nextflow distinguishes two different kinds of channels: queue channels and value channels.

  • A queue channel is an asynchronous unidirectional FIFO queue that connects two processes or operators. Channel elements are consumed till the channel is empty.
  • A value channel (a.k.a. a singleton channel) is bound to a single value and it can be read unlimited times without consuming its contents.

It is likely that one or more channel type has been converted to a queue channel and is only being read once.

In this example, the GATK_HAPLOTYPECALLER process is only executing one task.

You can see that the channel factory Channel.of has been used.

While parameters are treated as value channels, the Channel.of channel factory has converted these inputs to queue channels and are only being read once.

By removing these the inputs will be read multiple times.

Currently:

hello-gatk.nf
111
112
113
114
115
116
117
GATK_HAPLOTYPECALLER(
    SAMTOOLS_INDEX.out,
    Channel.of(params.genome_reference),
    Channel.of(params.genome_reference_index),
    Channel.of(params.genome_reference_dict),
    Channel.of(params.calling_intervals)
)

After: 

GATK_HAPLOTYPECALLER(
    SAMTOOLS_INDEX.out,
    params.genome_reference,
    params.genome_reference_index,
    params.genome_reference_dict,
    params.calling_intervals
)