jelena-tica/tiger

TIGER stands for Transduction Inference in GERmline genomes, and it performs mobile element mediated transduction detection.

TIGER

TIGER stands for Transduction Inference in GERmline genomes, and it performs non-reference mobile element mediated transduction detection.

The tool takes mobile element insertion predictions and translocation prediction calls and combines this information with the input from either a BWA or ELAND mapped BAMs files to generate a high- confidence transduction callset.

Installing TIGER

TIGER consists of a single bash script that mostly uses standard Unix commands. However, it relies on certain external tools that have to be pre-installed (see Dependencies) or are packaged into this repository for your convenience and to avoid problems with conflicting versions. Included are bamgrepreads, a tool from Pierre Lindenbaum, and a version of samtools.

Obtain this repository to your computer by either cloning it

git clone https://github.com/jelena-tica/TIGER.git

or downloading it as a zip file. Uncompress the zip in the latter case.

Enter the external directory

cd TIGER/external

and simply type

make

This will build samtools and bamgrepreads. Afterwards TIGER should be ready to run. You best execute the script with its full pathname, e.g.

/path/to/installation/TIGER/TIGER_v1.sh

Dependencies

You need the following external tools installed and in your PATH:

• Blat >= v34
• Bedtools (tested with version 2.17)
• Perl >= 5.8.x

To install blat, visit http://hgdownload.cse.ucsc.edu/admin/exe/, enter the folder corresponding to your machine type and download the blat pre-compiled binary (e.g. for Linux x86_64 machine, the following executable will be installed: http://hgdownload.cse.ucsc.edu/admin/exe/linux.x86_64/blat/blat)

To install bedtools visit https://github.com/arq5x/bedtools2 and download the current version (recommended). On a command line type:

tar -zxvf BEDTools-<version>.tar.gz
cd BEDTools-<version>
make

Run a test case

The folder test contains data from a small subset from an orangutan individual sequenced to 25x sequencing coverage in 2x100bp mode (Gokcumen et al., PNAS, 2013) that you can use to test your installation. This data is aligned with Eland, the non-reference mobile element calls come from a modified version of Transposable Element Analyzer - TEA (Lee et al., Science, 2012) and the translocation calls from Delly v0.0.11 - jumpy_v0.0.11 (Rausch et al., Bioinformatics, 2012). You still need the reference genome FASTA file for orangutan, which can be downloaded from the UCSC Genome Browser. It is recommended to run it on 20-30x sequencing coverage data, as it is currently extensively tested and used only on such data.

Then call TIGER like this (and replace <ponAbe2> with the Orangutan genome):

./TIGER_v1.sh test/test.bam eland <ponAbe2.fa> test/test.l1 test/test.trans test/

The results of such TIGER run can be found in expected_output.

Usage

When you call TIGER without arguments a usage synopsis will be shown:

TIGER_v1.sh bam aligner(bwa|eland) ref_genome L1_insertions translocations out_dir [min_match_len]

This input is required:

• bam: Whole-Genome Sequencing BAM file

• aligner: State which aligner has been used to map the BAM file from above (either eland or bwa)

• ref_genome: FASTA file containing the reference genome (e.g. from UCSC)

• L1_insertions: L1 insertion calls (from your favorite MEI caller) in the following TAB-separated format:

    chromosome_of_insertion	breakpoint1	breakpoint2	TSD
  

  where the breakpoint1 and the breakpoint2 are the start and end of a Target-Site Duplication (TSD) on a chromosome_of_insertion.

• translocations: Translocation calls (from your favorite SV caller) in the following TAB-separated format:

    chromosome1	breakpoint1	chromosome2	breakpoint2
  

  where the translocation happened between two locations in the genome. Both directions are taken into consideration (that the chromosome1 breakpoint1 translocates to chromosome2 breakpoint2 and vice versa).

• out_dir: Folder to write results to

• min_match_len (optional): Specify the minimum length of BLAT alignments to be considered. Defaults to 50. If you are using reads <=100bp, try lowering this value.

The final transduction file will have the following format:

TargetChr       TargetStart     TargetEnd       SourceChr       SourceStart     SourceEnd       SumOfReads      AverageUniq     ReadsWith6A/Ts  TSD     SourceSize 

• TargetChr, TargetStart, TargetEnd: Target chromosome, target start, target end (locus of insertion)
• SourceChr, SourceStart, SourceEnd: Source chromosome, source start, source end (locus of origin)
• SumOfReads: Sum of reads indicating the translocation which confirms transduction
• AverageUniq: The average "uniqueness" of reads indicating transduction (mean of how many times the read is found in the genome)
• ReadsWith6A/Ts: Number of reads containing at least six consecutive non-reference As/Ts
• TSD: Target Site Duplication (TSD) length
• SourceSize: Computationally assessed size of the predicted transduction

If you are running TIGER on multiple samples of the same species, it would be recommended to contruct the pool of all transduction calls (from each individual) and search your L1/MEI individual input callset for all of them, which will increase the sensitivity and improve the genotyping.

Filtering low-confidence regions

To make sure that TIGER produced high-confidence results, it is strongly advised to filter against presence of segmental duplications and known reference mobile elements. Since TIGER depends on MEI caller, raw non-reference MEI calls were filtered against known reference mobile elements based on indication coming from the MEI caller (in this case TEA specific oi=1 calls were considered as high-confidence since they do not overlap known reference repeats). Moreover only L1 calls were considered as we were investigating L1-mediated sequence transductions.

If your favorite MEI caller does not provide such filter, Repeat Masker files can be used. You can download them from the UCSC Genome Browser, by typing the following on the command line (shown here for the orangutan genome):

wget http://hgdownload.soe.ucsc.edu/goldenPath/ponAbe2/bigZips/chromOut.tar.gz
tar -zxvf chromOut.tar.gz

Please make sure to remove the header from each file (if separated by chromosome) or from one file (if there is only one file). If you are only looking at specific repeats, e.g. L1, you can first extract only L1. Once you have one Repeat Masker file containing all chromosomes without headers, you can contruct bed file by extracting query chromosome, position in query (begin) and position in query (end) or columns 5, 6 and 7 separated by TAB. Therefore you will now have a bed file and you can simply execute the following:

bedtools intersect -a <previous_TIGER_output.bed> -b <RepMask_file.bed> -v > TIGER_output_noRepMask.bed

Be careful to remove the header from the TIGER output file because bedtools will not run.

The same procedure can be done for segmental duplications if such files are available. The recommended TIGER run will filter the output against both known repeats (either by using the filter from your favorite MEI caller or by intersecting the calls against RepeatMasker given that your caller does not support such filter) and segmental duplications resulting in the list containing high-confidence transduction calls.