In cancer genomics, long-read de novo assembly approaches may not be applicable because of tumor heterogeneity, normal cell contamination and aneuploid chromosomes. Generating sufficiently high coverage for each derivative, potentially sub-clonal, chromosome is not feasible. Lorax is a targeted approach to reveal complex cancer genomic structures such as telomere fusions, templated insertions or chromothripsis rearrangements. Lorax is NOT a long-read SV caller, this functionality is implemented in delly.
Lorax is available as a statically linked binary, a singularity container (SIF file) or as a docker container. You can also build lorax from source using a recursive clone and make. Lorax depends on HTSlib and Boost.
git clone --recursive https://github.com/tobiasrausch/lorax.git
cd lorax/
make all
Lorax has several subcommands for alignments to linear reference genomes.
Templated insertions threads can be identified using
lorax tithreads -g hg38.fa -o tithreads.bed -m control.bam tumor.bam
The out.bed file specifies nodes (templated insertion source sequences) and edges (templated insertion adjacencies) of a graph that can be plotted using dot.
cut -f 4,9 out.bed | sed -e '1s/^/graph {\n/' | sed -e '$a}' > out.dot
dot -Tpdf out.dot -o out.pdf
The out.reads file lists unique assignments of reads to templated insertion source sequences. To extract the FASTA sequences for all these reads use the lorax extract subcommand (below) with the -a option.
tail -n +2 out.reads | cut -f 1 | sort | uniq > reads.lst
lorax extract -a -g hg38.fa -r reads.lst tumor.bam
Telomere-associated SVs can be identified with lorax using
lorax telomere -g t2t.fa -o outprefix tumor.bam
The output files cluster reads into distinct telomere junctions that can be locally assembled. Since telomeres are repetitive, common mis-mapping artifacts found in a panel of normal samples are provided in the maps subdirectory. It is recommended to use the telomere-to-telomere assembly as the reference genome for lorax telomere.
Given a list of amplicon regions and a phased VCF file, lorax can be used to extract amplicon reads for targeted assembly approaches.
lorax amplicon -g hg38.fa -s sample -v phased.bcf -b amplicons.bed tumor.bam
The amplicon subcommand outputs the selected reads (as a hash list out.reads) and a diagnostic table (out.bed) with amplicon regions and their support by split-reads. Ideally, all amplicon regions are connected and belong to one connected component (one cluster of amplicons). This amplicon graph can be plotted using dot.
cut -f 4,11 out.bed | sed -e '1s/^/graph {\n/' | sed -e '$a}' > out.dot
dot -Tpdf out.dot -o out.pdf
To extract the FASTA sequences for all reads use the lorax extract subcommand (below) with the -a option.
To get FASTA sequences and pairwise read to genome matches for a list of reads (list.reads) use
lorax extract -g hg38.fa -r list.reads tumor.bam
If the read list contains hashes instead of read names as from the lorax amplicon subcommand then please use the -a command-line option.
lorax extract -a -g hg38.fa -r list.reads tumor.bam
For pan-genome graphs and pan-genome graph alignments, lorax supports the below subcommands, some are work-in-progress.
lorax components pangenome.gfa.gz > comp.tsv
lorax gfa2dot -s s103 -r 3 pangenome.gfa.gz > graph.dot
dot -Tpng graph.dot > graph.png
With long reads aligned to a pan-genome graph
minigraph --vc -cx lr pangenome.gfa.gz input.fastq.gz | bgzip > sample.gaf.gz
lorax can be used to convert the graph alignment to BAM
lorax convert -g pangenome.gfa.gz -f input.fastq.gz sample.gaf.gz | samtools sort -o sample.bam -
lorax ncov -g pangenome.gfa.gz sample.gaf.gz > ncov.tsv
Tobias Rausch, Rene Snajder, Adrien Leger, Milena Simovic, Mădălina Giurgiu, Laura Villacorta, Anton G. Henssen, Stefan Fröhling, Oliver Stegle, Ewan Birney, Marc Jan Bonder, Aurelie Ernst, Jan O. Korbel
Long-read sequencing of diagnosis and post-therapy medulloblastoma reveals complex rearrangement patterns and epigenetic signatures
Cell Genomics, 2023, 100281, DOI: 10.1016/j.xgen.2023.100281
Lorax is distributed under the BSD 3-Clause license. Consult the accompanying LICENSE file for more details.