Tissue-specific variant annotation
Bioinformatics accepted version: http://academic.oup.com/bioinformatics/article/34/18/3061/4975416
This project aims at providing a functional annotation tool for genetic variation, using prior on different tissues. The general framework proposed with TiSAn will help researchers creating a predictive model for their tissue of interest. Here, two models are made available as examples: the human brain and heart.
Run the following command in R:
devtools::install_github("kevinVervier/TiSAn")
All packages should be installed automatically at the same time that TiSAn package.
User can find human brain and heart databases, as gzipped .bed files (+index) at: http://flamingo.psychiatry.uiowa.edu/TiSAn/
- TiSAn_Heart.bed.gz
- TiSAn_Heart.bed.gz.tbi
- TiSAn_Brain.bed.gz
- TiSAn_Brain.bed.gz.tbi
For the following examples, we assume that those databases have been downloaded and stored in the data repository.
To automatically annotate variants with TiSAn scores, we propose to use vcfanno tool (https://github.com/brentp/vcfanno). Please note that most of the annotation tools will work with a database in bed format.
The following command calls vcfanno on a VCF file containing positions to be annotated, and a configuration containing parameters for vcfanno.
vcfanno data/TiSAn.conf data/example1.vcf
The output is a new VCF file with an additional FORMAT column with two fields (TiSB for brain and TiSH for heart):
#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT
1 1005806 rs3934834 C T 0.0 PASS TiSB=0;TiSH=0.4307
1 243943084 rs4132509 C A 0.0 PASS TiSB=0;TiSH=0.5663
13 81753314 rs12584499 C G 0.0 PASS TiSB=0.718;TiSH=0
14 62763347 rs2354331 C T 0.0 PASS TiSB=0.358;TiSH=0.2904
21 37417489 rs2835248 A G 0.0 PASS TiSB=0;TiSH=0
4 154746806 rs10031057 A G 0.0 PASS TiSB=0;TiSH=0.8565
For region-based analysis, we provide genome-wide annotations on the UCSC Genome Browser website.
- Step 1: Access the custom track page:
- Step 2: Paste the following text into the "Paste URLs or data" box, and click "submit":
track type=bigWig name="Brain" description="TiSAn-Brain" visibility=full autoScale=off alwaysZero=on maxHeightPixels=100:30:10 color=24,181,84 bigDataUrl=http://flamingo.psychiatry.uiowa.edu/TiSAn/TiSAn_Brain.bw
track type=bigWig name="Heart" description="TiSAn-Heart" visibility=full autoScale=off alwaysZero=on maxHeightPixels=100:30:10 color=181,24,84 bigDataUrl=http://flamingo.psychiatry.uiowa.edu/TiSAn/TiSAn_Heart.bw
This will result in adding 2 tracks (TiSAn-Brain and TiSAn-Heart scores).
We also developed a GUI to efficiently annotate shortlists of genomic loci, and get details on individual feature used by TiSAn to make the predictions. Please refer to the TISAn-view folder on this Github repository.
We developed a GUI to efficiently create tissue-specific databases from publically available datasets. Different panels cover GTEx eQTLs, RoadMap Epigenomics DNA methylation, literature mining on PubMed, and a custom database that could be changed by users. Please refer to the TISAn-build folder on this Github repository.
