LinkSeq-Demux -- basecall/demultiplex and trim linked-reads

Table of Contents

• Overview
• Trimming
• Setup
• Running on tiny-bcl
  • Running demux pipeline
  • Output
• Barcode whitelist

Overview

This Nextflow pipeline basecalls and demultiplexes linked-reads from 10x Genomics and trims sequences, including barcode contamination. To run this pipeline, the 8-base sample indexes are needed, corresponding to the 10x Genomics indexes (e.g. SI-GA-A1).

This pipeline makes some assumptions about the input data. For example, it makes the assumption that it is paired-end sequencing, and therefore uses --use-bases-mask=Y*,I*,Y* in bcl2fastq, and assumes that the read lengths (and index length) is found in RunInfo.xml.

Trimming

There are four trimming steps in the nextflow pipeline, each of which is listed below.

• trim_adapters trims adapter sequences from 3' end using BBtools/BBDuk.
• bctrim trims linked-read barcodes from read 2. When the insert size is small and read 1 and 2 overlap, read 2 may be contaminated by the barcode attached to read 1. The bin/trimR2bc.py was taken from release v1.0 of trimR2bc by Elisabet Thomsen. See the GitHub repository for more details: https://github.com/ElisabetThomsen/trimR2bc
• polyG_trim trims poly-G tails from reads using fastp.
• quality_trim_read1/quality_trim_read2 trims low quality bases from read 1 and 2 respectively.

Setup

Install dependencies with conda using the environment.yml file:

conda env create -f environment.yml

Activate the environment (check the name of the environment, it should be linkseq-demux):

conda activate linkseq-demux

Pull this project with nextflow:

nextflow pull https://github.com/fargenfo/linkseq-demux

Running on tiny-bcl

Here's how to run this pipeline on the "tiny-bcl" example dataset from 10x Genomics. First of all, download the tiny-bcl tar file and the Illumina Experiment Manager sample sheet: tiny-bcl-samplesheet-2.1.0.csv.

Download the tiny-bcl data: https://support.10xgenomics.com/genome-exome/software/pipelines/latest/using/mkfastq#example_data

Next, edit the [Data] part of the samplesheet from the following:

Lane,Sample_ID,index,Sample_Project
5,Sample1,SI-GA-C5,tiny_bcl

To the following:

Lane,Sample_ID,index
5,Sample1,CGACTTGA
5,Sample1,TACAGACT
5,Sample1,ATTGCGTG
5,Sample1,GCGTACAC

Using that the index SI-GA-C5 corresponds to the four octamers CGACTTGA,TACAGACT,ATTGCGTG,GCGTACAC (10X Genomics have a tool for this on their website).

Also add adapter sequences to the [Settings] part of the samplesheet:

Adapter,AGATCGGAAGAGCACACGTCTGAACTCCAGTCA
AdapterRead2,AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT

Run demux pipeline

Use the tiny-bcl.config file to get an idea how to define input parameters. For the tiny-bcl dataset, you should set rundir to tiny-bcl-2.0.0 and samplesheet to tiny-bcl-samplesheet-2.1.0.csv. Additionally, you need the barcode whitelist, see the barcode whitelist section on how to obtain this list.

Change the memory and CPU specifications in the configuration (under process and executor) to suit your needs before continuing.

When you've made the config file and activated the linkseq-demux environment, run the pipeline like this:

nextflow run fargenfo/linkseq-demux -c [your config]

Output

The output from the pipeline, run on the tiny-bcl data, is shown below. The compressed FASTQ data is in outs/Sample1/fastqs. There are various logs, from the basecalling itself via Bcl2Fastq, from the various trimming steps, from read synchronization, and from FastQC. There is also a HTML report outs/multiqc/multiqc_report.html that combines the statistics from Bcl2Fastq with the FastQC report.

$ tree outs/
outs/
├── Sample1
│   ├── fastqc
│   │   ├── Sample1_L005_R1_fastqc.html
│   │   ├── Sample1_L005_R2_fastqc.html
│   │   ├── fastqc.log
│   │   └── zips
│   │       ├── Sample1_L005_R1_fastqc.zip
│   │       └── Sample1_L005_R2_fastqc.zip
│   ├── fastqs
│   │   ├── Sample1_L005_R1.fastq.gz
│   │   └── Sample1_L005_R2.fastq.gz
│   └── logs
│       ├── adapter_trim
│       │   └── L005.log
│       ├── bctrim
│       │   └── L005.log
│       ├── polyG_trim
│       │   └── L005.log
│       ├── quality_trim
│       │   ├── L005_R1.log
│       │   └── L005_R2.log
│       └── sync_reads
│           └── L005.log
├── bcl2fastq
│   ├── Stats.json
│   └── log.log
└── multiqc
    ├── multiqc_data
    │   ├── multiqc.log
    │   ├── multiqc_bcl2fastq_bylane.txt
    │   ├── multiqc_bcl2fastq_bysample.txt
    │   ├── multiqc_data.json
    │   ├── multiqc_fastqc.txt
    │   ├── multiqc_general_stats.txt
    │   ├── multiqc_qualimap_bamqc_genome_results.txt
    │   └── multiqc_sources.txt
    └── multiqc_report.html

Barcode whitelist

The whitelist contains barcodes in the 10X Genomics GemCode technology, and can be obtained in the LongRanger software bundle here:

longranger-2.y.z/longranger-cs/2.y.z/tenkit/lib/python/tenkit/barcodes/4M-with-alts-february-2016.txt

LongRanger can be obtained on the 10X Genomics website:

https://support.10xgenomics.com/genome-exome/software/overview/welcome

It's easier to download the file from this link though:

http://cb.csail.mit.edu/cb/ema/data/4M-with-alts-february-2016.txt