A software to detect virome-wide viral integrations
It is important to run "validate" function to remove possible false-positives to obtain the final list of candidate viral integrations.
Viral Integration caller (VIcaller) is a bioinformatics tool designed for identifying viral integration events using high-throughput sequencing (HTS) data. VIcaller is developed under Linux platform. It uses both FASTQ files or aligned BAM files as input. It also supports both single-end and paired-end reads. VIcaller contains one main Perl script, VIcaller.pl, that include three main functions: 1) detect, which will detect virome-wide candidate viruses and integration events; 2) validate, which will perform the in silico validation on those candidate viral integrations; 3) calculate, which will calculate the integration allele fraction. We also generated a comprehensive viral reference genome library with 411,195 unique whole and partial genomes, covering all six virus taxonomic classes. The viral reference genome library also comes with a taxonomy database in a defined format that give the virus name, etc.
VIcaller is an open-source software. VIcaller.v1.1 source code is also available at http://www.uvm.edu/genomics/software/VIcaller.html. You need to get the virome-wide library and vector database at http://www.uvm.edu/genomics/software/VIcaller.html.
$ tar vxzf VIcaller.tar.gz
$ cd VIcaller/
$ mkdir Tools
a) Currently VIcaller relies on the following dependencies to be compiled (contact Dr. Xun Chen if you need help get those tools or Perl libraries installed).
b) Obtain the installed file from the following links.
c) Follow the instruction to successfully install each tool (contact server manager if there is any compile issues).
d) Check or install the listed Perl libraries using cpan, cpanm or other methods.
• BWA (default version: v0.7.10): https://github.com/lh3/bwa/tree/master/bwakit
• Bowtie2 (default version: v2.2.7): https://sourceforge.net/projects/bowtie-bio/files/bowtie2/2.2.7/
• TopHat2 (v2.1.1): http://ccb.jhu.edu/software/tophat/index.shtml
• BLAT (default version: v.35): http://genomic-identity.wikidot.com/install-blat
• BLAST+ (default version: v2.2.30): http://mirrors.vbi.vt.edu/mirrors/ftp.ncbi.nih.gov/blast/executables/blast%2B/2.2.30/
• SAMtools (default version: v1.6): https://sourceforge.net/projects/samtools/
• HYDRA (default version: 0.5.3): https://code.google.com/archive/p/hydra-sv/downloads
• NGS QC Toolkit (default version: v2.3.3): the installer can be found under the VIcaller/Tools/ folder
a) Copy the script “TrimmingReads_sanger.pl” under the VIcaller/Scripts/ folder to the installed NGSQCToolkit_v2.3.3/Trimming/ folder
• FastUniq (Default version: v1.1): https://sourceforge.net/projects/fastuniq/
• SE-MEI (modified): https://github.com/dpryan79/SE-MEI (original version), the modified version can be found under the VIcaller/Tools/ folder
a) Copy the modified SE-MEI installer (SE-MEI-master.tar.gz) under the VIcaller/Scripts/ folder to the VIcaller/Tools/ folder
b) Install the modified SE-MEI tool follow the README file
• RepeatMasker (default version: v4.0.5):
a) Install RepeatMasker: http://www.repeatmasker.org/
b) Install RMBlast aligner: http://www.repeatmasker.org/RMBlast.html
c) Compile the Repbase database: https://www.girinst.org/repbase/
• MEME (default version: v4.11.1): http://web.mit.edu/meme_v4.11.4/share/doc/download.html
• TRF (default version: v4.07b): https://tandem.bu.edu/trf/trf.html
$ cpan String::Approx
$ cpan Time::HiRes
$ cpan Test::Most
$ cpan Bio::Seq
$ cpan Bio::SeqIO
$ cpan Bio::DB::GenBank
$ cpan IO::Zlib
$ cd VIcaller/Database/Human/
$ wget http://hgdownload.soe.ucsc.edu/goldenPath/hg38/bigZips/hg38.fa.gz
$ gunzip hg38.fa.gz
$ bwa index -a bwtsw hg38.fa
$ bowtie2-build hg38.fa hg38.fa
$ makeblastdb -in hg38.fa -dbtype nucl
$ cd VIcaller/Database/Virus/
$ bwa index -a bwtsw virus_db_090217.fa
$ bowtie2-build virus_db_090217.fa virus_db_090217.fa
$ makeblastdb -in virus_db_090217.fa -dbtype nucl
export PERL5LIB=/users/xchen/.cpan/build/
export PATH=$PATH:/users/xchen/VIcaller/Tools/bowtie2-2.2.7/
# human_genome = /users/xchen/VIcaller/Database/Human/hg38.fa
# human_genome_tophat = /users/xchen/VIcaller/Database/hg38.fa
# virus_genome = /users/xchen/VIcaller/Database/Virus/virus_db_090217.fa
# virus_taxonomy = /users/xchen/VIcaller/Database/Virus/virus_db_090217.taxonomy
# virus_list = /users/xchen/VIcaller/Database/Virus/virus_db_090217.virus_list
# vector_db = /gpfs2/dli5lab/CAVirus/Database/Vector/Vector.fa
# cell_line = /users/xchen/VIcaller/Database/cell_line.list
# bowtie_d = /users/xchen/VIcaller/Tools/bowtie2-2.2.7/
# tophat_d = /users/xchen/VIcaller/Tools/tophat-2.1.1.Linux_x86_64/
# bwa_d = /users/xchen/VIcaller/Tools/bwa-master/
# samtools_d = /users/xchen/VIcaller/Tools/samtools-1.6/
# repeatmasker_d = /users/xchen/VIcaller/Tools/RepeatMasker/
# meme_d = /users/xchen/VIcaller/Tools/meme_4.11.1/
# NGSQCToolkit_d = /users/xchen/VIcaller/Tools/NGSQCToolkit_v2.3.3/
# fastuniq_d = /users/xchen/VIcaller/Tools/FastUniq/
# SE_MEI_d = /users/xchen/VIcaller/Tools/SE-MEI/
# hydra_d = /users/xchen/VIcaller/Tools/Hydra-Version-0.5.3/
# blat_d = /users/xchen/bin/x86_64/
# blastn_d = /users/xchen/VIcaller/Tools/ncbi-blast-2.2.30+-src/bin/
#. Make sure the space between “#” and parameters.
#. Make sure the directory for the Perl library is correct or the libraries are available in the path if you install them locally.
#. Make sure the Bowtie2 directory is correct or it is available in the path (recommended) if you are going to analyze RNA-seq data.
#. Make sure the human and virus databases existed and correctly indexed.
$ perl VIcaller.pl <functions> [arguments]
$ perl VIcaller.pl detect [arguments]
a) WGS data in single-end fastq format:
$ perl VIcaller.pl detect -d WGS -i seq -f .fastq.gz -s single-end -t 12
b) RNA data in paired-end fastq format (set bowtie2 path before run the following command):
$ perl VIcaller.pl detect -d RNA-seq -i seq -f .fastq.gz -s paired-end -t 12
c) RNA alignment data in bam format (Note: Human reference genome should be the same as the bam file)
$ perl VIcaller.pl detect -d RNA-seq -i seq -f .bam -s paired-end -t 12
4.2 Validate candidate viral integrations (will remove false viral integrations detected at the "detect" step due to reads mis-alignment)
$ perl VIcaller.pl validate [arguments]
$ perl VIcaller.pl validate -i seq -S seq_1_24020575_24020787_HPV16_218931404 -G 218931404 -V HPV16
$ perl VIcaller.pl calculate [arguments]
$ perl VIcaller.pl calculate -i seq -f .fastq.gz -S -C 1 -P 24020575 -B 2 -N 20
The candidate viral integrations detected by VIcaller are kept in the file with suffix of “.output” in Viral integration Format (VIF), with the visualization of the aligned read sequences in the file with suffix of “.visualization”. After in silico validation and allele fraction calculation, the results are also kept in the output file. “seq” is an example sample ID.
Details are included in the Manual PDF file
The following Linux command can be used to extract the information required to run human genome functional annotation tools. The VIcaller output file is “seq.output”, and for example, if the functional annotation software is SnpEff, the following command line will extract the information required to run SnpEff. The output from using this command will be the input file for SnpEff.
$ awk '{if ($7!="Chr.")print$7"\t"$17"\t.\tA\tT\t."}' seq.output
“Fast” mode is significantly faster than “Standard” mode. However, the “Fast” mode does not analyze viral reads, which are supporting evidence for distinguishing between viral integrations and viral infections.
VIcaller analyzes individual samples and then generates a list of viral integrations for each sample. Viral integration enrichment (bias) analysis, which is a statistical analysis, requires inclusion of a group of samples. The enrichment analysis has to be performed solely. There are multiple statistical models to calculate/determine enrichment hotspots (such as simulation-based Z score test). There are many available tools and R packages that can be selected for enrichment analysis. Users may have different preference on statistical models to fit in their actual samples/data.
6.4 Can I use the published tools that were designed for detecting transposable element insertions to identify virome-wide integrations?
VIcaller uses the reads that are commonly used in transposable element insertion and other structural variation detection tools. However, because VIcaller is specifically designed to identify virome-wide integrations, it has significant advantages over alignment-based transposable element insertion detection tools for viral integration analysis, which are designed to extract and mainly use (human’s) anomalous reads specifically. For example, 1) VIcaller supports the use of virome-wide library as the reference to detect any characterized viruses, while most transposable element detection tools use transposable element sequences as the reference; and 2) VIcaller implements viral integration-specific quality control procedures and implements additional steps to in silico verify detected viral integrations. We have tried to compare VIcaller with other transposable element insertion detection software, e.g., MELT. MELT failed to run in a virome-wide fashion after we replaced MELT’s default consensus transposable element reference sequences with our virome-wide database. We further tested whether MELT was able to detect simulated candidate viral integrations, and we found that although MELT did run, it was not able to detect any of these integrations.
You can use other viral databases as the reference. However, the final output may not include the viral names or other taxonomy information. The reads that multiple-mapped different viral sequences from the same virus may not be efficiently recovered for the detection of viral integrations.
We previously used hg19 reference build to generate the test output file and to prepare the input BAM files. Thus, you should use hg19 to run the test data in BAM format or if you want to compare with the example output file.
Xun Chen, Jason Kost, Arvis Sulovari, Nathalie Wong, Winnie S. Liang, Jian Cao, and Dawei Li. A virome-wide clonal integration analysis platform for discovering cancer viral etiologies. Genome Research. 2019 Mar. DOI: 10.1101/gr.242529.118
http://www.uvm.edu/genomics/software/VIcaller.html
VIcaller is licensed under the Creative Commons Attribution-NonCommercial 4.0 International license. It may be used for non-commercial use only. For inquiries about a commercial license, please contact the corresponding author at [email protected] or The University of Vermont Innovations at [email protected].