Skip to content

mskcc/nf-fastq-plus

BranchesTags

Repository files navigation

nf-fastq-plus

Generate IGO fastqs, bams, stats and fingerprinting

Run

There are two options for running the modules in this pipeline -

Links for Developers

Demultiplex and Stats

Description: Runs end-to-end pipeline of demultiplexing and stats. The input of this is the name of the sequencing run

# Basic
nextflow main.nf --run ${RUN}

# Skip demultiplexing
nextflow main.nf --run ${RUN} --force true

# Run demux and stats only on one request
nextflow main.nf --run ${RUN} --filter ${REUQEST_ID}

# Run in background
nohup nextflow main.nf --run ${RUN} --force true -bg

# Push pipeline updates to nf-dashboard
nohup nextflow main.nf --run ${RUN} --force true -with-weblog 'http://dlviigoweb1:4500/api/nextflow/receive-nextflow-event' -bg

Arguments (--arg)

  • --run: string (required), directory name of the sequencing directory

    Eg: 210406_JOHNSAWYERS_0277_000000000-G7H54

  • --force: string (optional), skips the demultiplexing if already completed

    Eg: true, false

  • --filter: string (optional), only run requests that match this filter

    Eg: 10000_B (only runs samples in 10000_B request), 10000 (only runs samples in 10000 request, i.e. NOT 10000_B samples)

Options (-opt)

  • -bg: run process in background
  • -with-weblog: publish events to an API

Stats Only

Description: Runs stats given a demultiplex output

# Basic
nextflow samplesheet_stats_main.nf --ss ${SAMPLE_SHEET} --dir ${DEMULTIPLEX_DIRECTORY} 

# Run stats only on one request
samplesheet_stats_main.nf --ss ${SAMPLE_SHEET} --dir ${DEMULTIPLEX_DIRECTORY}  --filter ${REUQEST_ID}

# Run in background
nohup nextflow samplesheet_stats_main.nf --ss ${SAMPLE_SHEET} --dir ${DEMULTIPLEX_DIRECTORY} -bg

Arguments (--arg)

  • --dir: string (required), Absolute path to the directory name of the demultiplexed directory

    Eg: /igo/work/FASTQ/DIANA_0333_BH53GNDRXY_i7

  • --ss: string (required), Absolute path to the sample sheet that CREATED the value of --dir

    Eg: /home/igo/DividedSampleSheets/SampleSheet_210407_DIANA_0333_BH53GNDRXY_i7.csv

  • --filter: string (optional), only run requests that match this filter

    Eg: 10000_B (only runs samples in 10000_B request), 10000 (only runs samples in 10000 request, i.e. NOT 10000_B samples)

Options (-opt)

  • -bg: run process in background

Re-running Pipeline

  • Demultiplexing will fail if the FASTQ directory already exists.
    • If demultiplexing is required, remove the FASTQ directory
    • If demultiplexing can be skipped, add the --force true option
  • Stats will be skipped if the final BAM for that sample has been written to ${STATS_DIR}/${RUNNAME}/${RUN_TAG}.bam
    • If stats need to be re-run, remove relevant BAMs from the ${STATS_DIR} folder specified in nextflow.config

For Development

Please Read:

  • Create a feature/{YOUR_CHANGE} branch for new features or hotfix/{YOUR_FIX} for future development
  • Before merging your branch into master, wait for the GitHub actions to run and verify that all checks pass. Do not merge changes if there are failed tests. Either talk to IGO Data Team or fix the tests.

Project Structure

  • Follow the project structure below -
.
├── README.md
├── main.nf
├── modules
│   └── m1.nf
├── nextflow.config
└── templates
    └── process.sh
  • templates: Where all scripts (bash, python, etc.) will go. Don't rename this directory because nextflow is seutp to look for a directory of this name where the nextflow script is run
  • modules: Directory containing nextflow modules that can be imported into main.nf

Adding a new workflow

  • Passing sample-specific parameters (e.g. Reference Genome, Recipe, etc.) is done via a params file w/ key=value space-delimited values. To use this file, make sure that a {PREVIOUS_WORKFLOW}.out.PARAMS file is passed to the workflow and specified as a path-type channel. Make sure to use the .out.PARAMS of the workflow that the next_wkflw should be dependent on. I've noticed that nextflow won't pass all outputs of a workflow together (e.g. BAM of one task and the run params folder of another task)

Steps for Adding a New Module

  1. Add module
├── modules
│   └── process.nf

process {PROCESS_NAME} {
  [ directives ]

  output:
  ...
  stdout()

  shell:
  template '{PROCESS_SCRIPT}'
}
  • You don't need to import the template script. From the documentation, "Nextflow looks for the template file in the directory templates that must exist in the same folder where the Nextflow script file is located"
  • Note: Add the stdout() as an output if you would like to log the out to the configured log file
  1. Add template
└── templates
    └── process.sh
  • Write whatever script with the appropriate header (e.g. #!/bin/bash) that includes the following
    • Nextflow Inputs: Inputs defined as nextflow Input values. Add (Input) if defined in process Input or (Config) if defined in `nextflow.config
    • Nextflow Outputs: Outputs that will be defined in the execution context
    • Run: A sample command of how to run the script
#!/bin/bash
# Submits demultiplexing jobs
# Nextflow Inputs:
#   RUN_TO_DEMUX_DIR: Absolute path to directory of the Run to demux (Defined as input in nextflow process)
# Nextflow Outputs:
#   DEMUXED_RUN: Name of run to demux, given by the name of @RUN_TO_DEMUX_DIR
# Run:
#   RUN_TO_DEMUX_DIR=/igo/sequencers/michelle/200814_MICHELLE_0249_AHMNCJDRXX ./demultiplex.sh
  1. Emit PARAM file (Only if downstream processes are dependent on the output)
  • If your process is a dependency of downstream processes, emit the PARAMS file (nextflow.config parameter RUN_PARAMS_FILE) so that it can be read directly by the receiving channel along w/ the process's value.
process task {
  ...

  input:
  path PARAMS
  path INPUT
 
  output:
  path "${RUN_PARAMS_FILE}", emit: PARAMS       # Emit the same params value passed into the task
  path '*.bam', emit: VALUE

  shell:
  template 'task.sh'
}

workflow wkflw {
  take:
    PARAMS
    INPUT

  main:
    task( PARAMS, INPUT )
  
  emit:
    PARAMS = task.out.PARAMS                    # Assign PARAMS so that it's available in the main.nf
    VALUE = task.out.VALUE
}
  • Why? Nextflow channels emit asynchronously. This means that upstream processes will emit and pass to the next available process and not necessarily the expected one. For instance, if process A emits parameters used by all downstream processes and process B emits the value that will be transformed by that parameter, process C will not necessarily receive the proccess A parameters that apply to value emited by process B because each process has an independent, asynchronous channel.
  1. (Optional) Add logging

In the modules, convert the exported member to a workflow that calls an included log_out process to log everything sent to stdout by the process. See below,

include log_out as out from './log_out'

process task {
  output:
  stdout()		// Add this to your outputs
  ...

  shell:
  '''
  echo "Hello World" 	// Example sent to STD OUT
  ...
  '''
}

workflow task_wkflw { 	// This is what will actually be exported
  main:
    task | out
}

Logging

There are three files that log information -

  • LOG_FILE: All output is logged here (except commands)
  • CMD_FILE: All stat commands are logged to this file
  • DEMUX_LOG_FILE: All demultiplexing commands are logged here

Testing

Docker Container Actions run our integration tests on GitHub. To test changes, please build the dockerfile from the root and verify no errors are generated from the samplesheet_stats_main_test_hwg.sh and cellranger_demux_stats.sh scripts.

docker image build -t nf-fastq-plus-playground .

# Test stats-only workflow
docker run --entrypoint /nf-fastq-plus/testPipeline/e2e/samplesheet_stats_main_test_hwg.sh -v $(pwd)/../nf-fastq-plus:/nf-fastq-plus nf-fastq-plus-playground

# Test e2e (demux & stats)
docker run --entrypoint /nf-fastq-plus/testPipeline/e2e/cellranger_demux_stats.sh -v $(pwd)/../nf-fastq-plus:/nf-fastq-plus nf-fastq-plus-playground

Nextflow Config

Modify directory locations, binaries, etc. in the nextflow.config file

Important Files

LOG_FILE        # Logs all output from the pipeline
CMD_FILE        # Logs all commands from the pipeline (e.g. was bcl2fastq run w/ 1 or 0 mistmaches?)
DEMUX_LOG_FILE  # Logs output of bcl2fastq commands

Important Directories

STATS_DIR                   # Where final BAMS are written to
STATSDONEDIR                # Where stat (.txt) files & cellranger ouptut is written to
PROCESSED_SAMPLE_SHEET_DIR  # Where split samplesheets go (these are used for demuxing and stats)
LAB_SAMPLE_SHEET_DIR        # Original source of samplesheets
COPIED_SAMPLE_SHEET_DIR     # Where original samplesheets are copied to
CROSSCHECK_DIR              # Directory used for fingerprinting
SHARED_SINGLE_CELL_DIR      # Directory used by DLP process to create metadata.yaml (should happen automatically)

Other

LOCAL_MEM                   # GB of memory to give a process (e.g. demultiplexing) if executor=local

Crontab Setup

The pipeline can be kicked off automatically by the crontab/detect_copied_sequencers.sh script. Add the following to enable the crontab

# crontab -e
SHELL=/bin/bash

# Add path to bsub executable
PATH=${PATH}:/igoadmin/lsfigo/lsf10/10.1/linux3.10-glibc2.17-x86_64/bin

# Load the LSF profile prior to running the command
* * * * * . /igoadmin/lsfigo/lsf10/conf/profile.lsf; lsload; bhosts; /PATH/TO/detect_copied_sequencers.sh >> /PATH/TO/nf-fastq-plus.log 2>&1

Docker Container

  • v1.0.0 - First Release
  • v1.0.1 - Testing against "ncbi-genomes-2021-09-23" for GRCh38.p13
  • v1.0.2 - Testing against "GRCh38.p13.dna.primary.assembly.fa" for GRCh38.p13

Note - Changes to the github actions using the docker file need to be tagged, e.g.

VERSION_NUMBER=...      # e.g. "v2"
git add ...
git commit -m "My change"
git tag -a -m "very important change" ${VERSION_NUMBER} 
git push --follow-tags

REF

Debug

Demultiplexing

I) Samplesheet index length doesn't match RunInfo.xml length

BCL2FASTQ

  • Use use-bases-mask

DRAGEN
ERROR: Sample Sheet Error: SampleSheet.csv sample #1 (index 'GGTGAACC') has an index of length 8 bases, but a length of 10 was expected based upon RunInfo.xml or the OverrideCycles setting. Solution: Mask the indices by adding the OverrideCycles option to the SampleSheet

```
[Settings],,,,,,,,
OverrideCycles,Y151;I8N2;I8N2;Y151
,,,,,,,,
```

II) Only the ___MD.txt file is available in the STATSDONEDIR for a sample

  • Delete the "final sample BAM" in STATSDONEDIR, e.g. 
    RUN_NAME=...        # ROSALIND_0001_AHNKNJDSX2
SAMPLE_NAME=....    # 01001_A_1
rm /igo/staging/stats/${RUN_NAME}/*${SAMPLE_NAME}*.ba*
  • Why? The final sample bam was created, but the stats for that BAM did not finish or were interrupted.
    • This can happen when the pipeline is interrupted after writing the "final" sample .bam to SAMPLE_BAM_DIR. The generate_run_params_wkflw checks whether this .bam file is written
    • Issue to address this nf-fastq-plus issue 202

III) My github action change isn't reflecting in the integrated tests

About

Generate IGO fastqs, bams, stats and fingerprinting

Resources

Readme
Activity
Custom properties

Stars

1 star

Watchers

7 watching

Forks

0 forks
Report repository

Releases

3 tags

Packages

No packages published

Contributors 4

  •  
  •  
  •  
  •  

Languages