Motivation

• Association of enhancers and promoter regions with specifically expressed alleles

• Unambiguous elucidation of both cis- and trans- eQTLs

Aims

Workflow

1. Contruct spliced variant graphs

1a. Single whole genome graph

1b. Multiple disjoint gene (splice) graphs

2. Map short reads to graph(s)

3. Surject mapped reads to haplotype-specific transcripts

4. Run RSEM or similar

Implementation

Download the relevant pipeline scripts

wget https://raw.githubusercontent.com/NCBI-Hackathons/TheHumanPangenome/master/RNA/wdl_pipeline/vg_rna.wdl -O vg_rna.wdl
wget https://raw.githubusercontent.com/NCBI-Hackathons/TheHumanPangenome/master/RNA/wdl_pipeline/vg_rna.inputs.json -O vg_rna.inputs.json

Download cromwell (requires java 8 and Docker to run)

Quickstart intro on using Cromwell: (https://cromwell.readthedocs.io/en/develop/tutorials/FiveMinuteIntro/)

wget https://github.com/broadinstitute/cromwell/releases/download/36.1/cromwell-36.1.jar .

Download example input files

wget ftp://ftp.ensembl.org/pub/release-94/fasta/homo_sapiens/dna/Homo_sapiens.GRCh38.dna.primary_assembly.fa.gz .
bgzip -d Homo_sapiens.GRCh38.dna.primary_assembly.fa.gz
samtools faidx Homo_sapiens.GRCh38.dna.primary_assembly.fa

wget ...

Run the WDL workflow

sudo java -jar cromwell-36.1.jar run vg_rna.wdl -i vg_rna.inputs.json

Results

*Note that the numbers shown above for the simulated data (first two rows) are for alignments that have not been filtered according to mapping quality

Future goals

• Complete allele-specific expression pipeline

• Comprehensive evaluation of splice graph mapping using vg