FusionHunt: A Fusion Gene Detection Software

Bioinformatics Data Integration Practicum Project - Team 2

Description

This software, the FusionHunt, is a pipeline that detects possible fusion genes based on de novo transcriptome assembly from human RNA-seq data with benchmark result and functionality as existing software GFusion for clean and efficient fusion gene detection.

The pipeline runs by first assembling the reads by de novo transcriptome assembly using Trinity, removing tandem repeats and interspersed repeats using Tandem Repeat Finder and RepeatMasker, then aligning to the genome using BLAT, detecting possible chimeric transcripts using R-SAP, and filtering out for more likely fusion genes by aligning back to the original reads using Bowtie. Then, the pipeline will compare its results directly to those from GFusion, a pipeline that uses reference-based assembly to detect fusion genes.

Prerequisites

• Linux environment with bash
• Trinity v2.8.4 or higher with trimmomatic embedded
• Tophat v2.1.1 or higher
• Bowtie v1.1.2
• Bowtie2 v2.2.7
• Samtools v1.3.1
• Jellyfish v2.2.6 or higher
• Salmon v0.9.1
• GCC 8.2.0 or higher
• Tandem Repeat Finder v4.09 or higher
• RepeatMasker v1.332 or higher
• RMBlast v2.60+ or higher
• ActivePerl v5.8.0 or higher
• Perl Bioperl module v1.007 or higher
• R-SAP v1.1 or higher
• BLAT
• GFusion v1.0 (optional)
• RSEM v1.1.17 or higher (optional)

Configuration

Although some software being used in this pipeline are by default installed and added to $PATH, some require manual specification of path information.

Please make sure that the following is true

• Trinity is accessible via Trinity command
• Tophat is accessible via tophat command
• Bowtie is accessible via bowtie command
• Samtools is accessible via samtools command
• RSEM is accessible in command line directly (optional)
• ActivePerl with BioPerl module installed is accessible via perl command. (You may need to add ActivePerl to your PATH in order to do that)
• Trimmomatic installation directory is accessible via the $TRIMMOMATIC_DIR environment variable

Then, please fill the config.ini file with the installation paths of the remaining software in order to get the pipeline working as intended.

Usage

You should call main.sh with the following command line arguments:

Usage: ./main.sh -l|--left <left_seq> -r|--right <right_seq> -g|--genome <genome_seq> -c|--coord <ref_coord> [-h|--help] [-v|--version] [-o|--output-dir <output_dir>] [--cpu <number_of_cores>] [--max-memory <max_memory>] [--trimmomatic-param <param>] [--GFusiion] [--GFusion-bowtie-index <bowtie_index_prefix>] [--GFusion-opt <gfusion_options>] [--rsem] [--rsem-opt <rsem_options>]

  Example:
    ./main.sh -l left.fq -r right.fq -g genome.fasta -c coord.gtf --cpu 8 --max-memory 10G --GFusion --GFusion-bowtie-index data/index/hg31

  Required:
    -l|--left <left_seq>:         The left sequence
    -r|--right <right_seq>:       The right sequence
    -g|--genome <genome_seq>:     The reference genome sequence
    -c|--coord <ref_coord>:       The reference transcript coordinate file in
                                  GTF format or zipped GTF files

  Optional:
    -h|--help:                    Show the help information for this pipeline
    -v|--version:                 Show the version information for this pipeline
    -o|--output-dir <output_dir>: The output directory (default is output/)
    --cpu <number_of_cores>:      The number of CPU cores to work (default is
                                  total number of cores minus 2)
    --max-memory <max_memory>:    The max number of memory to be used (default
                                  is half of the total memory)
    --GFusion:                    Perform genome alignment-based fusion gene
                                  detection powered by GFusion (default is
                                  disabled)
    --rsem:                       Perform RSEM estimation of gene and isoform
                                  expression (default is disabled)

  Subroutine-specific Optional Parameters:
    --trimmomatic-param <param>:  Adapters and specifications for trimmomatic
                                  to perform automatic quality trim (default
                                  is "ILLUMINACLIP:$TRIMMOMATIC_DIR/
                                  adapters/TruSeq3-PE.fa:2:30:12")
    --GFusion-bowtie-index <idx>: The path and prefix of bowtie index file.
                                  NOTICE: This is REQUIRED if GFusion is enabled.
    --GFusion-opt <options>:      Optional arguments passed into GFusion.
                                  'options' must be en entire string such
                                  as "-r 0 -n 1"
    --rsem-opt <options>:         Optional arguments passed into RSEM.
                                  'options' must be en entire string such
                                  as "-r 0 -n 1"

Output

The output will be in <output_dir>/rsap_files/ChimericTranscriptAnnotation.out.

FASTQ File formatting

Please use a file format that is supported by Trinity. If you are downloading data from SRA, make sure to use this command: fastq-dump --defline-seq '@$sn[_$rn]/$ri' --split-files <accession_ID>

Genome downloads:

• Human Genome (hg38/GRCh38) from UCSC in fasta/fa format can be downloaded here for the -g flag
• Human Genome (hg38/GRch38) transcript file in gtf format can be downloaded here for the -c flag

Known bugs and issues

• Bugs from prerequisite software:

There is a bug in Trinity v2.2.8-v2.3.4.1 that puts the entire program into an infinite loop. Please use Trinity v2.2.7

• Bugs within this pipeline:

There are currently no known bugs at this point. However, should there be any issues please submit it to the issues page.

The group

• Program Manager: Sarah Hsu
• Technical Lead: Siwei Xu
• Lead Technical Writer: Martin Ma
• Q&A Lead: Juntao Chen
• Communications Lead: Saina Mahera