InfoGenomeR

InfoGenomeR is the Integrative Framework for Genome Reconstruction that uses a breakpoint graph to model the connectivity among genomic segments at the genome-wide scale. InfoGenomeR integrates cancer purity and ploidy, total CNAs, allele-specific CNAs, and haplotype information to identify the optimal breakpoint graph representing cancer genomes.

Requirements

bwa
samtools
bedtools
blat
R
lpSolveApi
ABSOLUTE
BICseq2

Inputs

genome-binning read depths (cn_norm, cn_norm_germ)
Initial SV calls (delly.format, manta.format, novobreak.format)
Non-properly paired reads (NPE.fq1, NPE.fq2)
SNP calls (het_snps.format, hom_snps.format)

Outputs

Haplotype-resolved SVs and CNAs (SVs.CN_opt.phased, copy_number.CN_opt.phased)
Haplotype graph (node_keys, edge_information.txt)
Karyotypes (Eulerian_path.0)
SV cluster and topology (cluster_sv)

How to generate inputs from BAM

We provide scripts for generating inputs using BIC-seq2, DELLY2, Manta, and novobreak, but we recommend that a user follow the guideline of each tool.
Generate cn_norm and cn_norm_germ.
- Install BICseq2 and modifiedSamtools.
- Set configs in the script bicseq_preprocess.sh and run the script for cn_norm.
- Set configs in the script bicseq_preprocess_germ.sh and run the script for cn_norm_germ.

norm_script=./NBICseq-norm_v0.2.4/NBICseq-norm.pl
map_file=./NBICseq-norm_v0.2.4/hg19.CRG.50bp/
read_length=100
fragment_size=350
tumor_bam=tumor.bam

./preprocessing/bicseq_preprocess.sh

norm_script=./NBICseq-norm_v0.2.4/NBICseq-norm.pl
map_file=./NBICseq-norm_v0.2.4/hg19.CRG.50bp/
read_length=100
fragment_size=350
normal_bam=normal.bam

./preprocessing/bicseq_preprocess_germ.sh

Generate NPE.fq1 and NPE.fq2

samtools view -h -F 2 -b simulated_sorted.bam >simulated_sorted_NPE.bam
samtools sort -n simulated_sorted_NPE.bam > simulated_sorted_NPE_sorted.bam
bamToFastq  -i simulated_sorted_NPE_sorted.bam  -fq NPE.fq1 -fq2 NPE.fq2

Generate delly SV calls (delly.format).
- Install DELLY2, set configs, and run the script.

ref=hg19.fa
tsv=samples.tsv
tumor_bam=tumor.bam
normal_bam=normal.bam

./preprocessing/delly_somatic.sh 3 20

Generate manta SV calls (manta.format).
- Install Manta, set configs, and run the script.

reference=hg19.fa
script1=./manta-1.1.0.centos5_x86_64/bin/configManta.py
script2=./runWorkflow.py
normal_bam=normal.bam
tumor_bam=tumor.bam

./preprocessing/manta_somatic.sh

Generate novobreak SV calls (novobreak.format).
- Install novobreak and copy the scripts ./preprocessing/run_novoBreak_orient_partial_corrected.sh and ./preprocessing/infer_sv_orientation.pl to the novobreak-installed folder.
- Set configs, and run the script.

novo_dir=./novoBreak_distribution_v1.1.3rc
reference=hg19.fa
tumor_bam=tumor.bam
normal_bam=normal.bam

./preprocessing/novobreak_somatic.sh

Generate SNP calls (het_snps.format and hom_snps.format).
- Set configs and run the script.

reference=hg19.fa
normal_bam=normal.bam
tumor_bam=tumor.bam

./preprocessing/het_SNP_detection_somatic.sh

Running InfoGenomeR

Breakpoint graph construction.
./breakpoint_graph/breakpoint_graph.sh <mode> <sample_name> <cancer_type> <min_ploidy> <max_ploidy> <bicseq_norm> <bicseq_norm_germ> <copy_numbers.control> <fasta_prefix> <haplotype_coverage> <tumor_bam> <normal_bam> <chr_prefix>
Allele-specific graph construction.
./allele_graph/allele_graph.sh <mode> <copy_numbers.control> <hom_snps.format> <het_snps.format> <fasta>
Haplotype graph construction.
./haplotype_graph/haplotype_graph.sh

Demos

Download demo files. Demo contains input files for InfoGenomeR.
Demo 1: a simiulated cancer genome (haplotype coverage 5X, triploidy, purity 75%) that has 162 somatic SVs (true_SV_sets_somatic). Demo 1.
Demo 2: A549 cancer cell line that is triploidy with der(11)t(8;11) and der(19)t(8;19)x2. It has chromothripsis on chromosome 15. Demo 2. From initial 4054 SV calls, InfoGenomeR reconstructs der(11)t(8;11) and der(19)t(8;19)x2.

Demo 1

Check baselines for SVs.
Rscript SV_performance.R delly.format true_SV_sets_somatic 0 0
precision: 0.6902174 recall: 0.7987421 fmeasure: 0.7405248
Rscript SV_performance.R manta.format true_SV_sets_somatic 0 0
precision: 0.9495798 recall: 0.7106918 fmeasure: 0.8129496
Rscript SV_performance.R novobreak.format true_SV_sets_somatic 0 0
precision: 0.9021739 recall: 0.4968553 fmeasure: 0.6408014
Running InfoGenomeR.
Merge SV calls.
cat delly.format manta.format novobreak.format > SVs
Run scripts for breakpoint graph construction.
./breapkpoint_graph/breakpoint_graph.sh somatic sample1 sample1 2 4 bicseq_norm bicseq_norm_germ copy_numbers.control hg19 5 null null 0
./allele_graph/allele_graph.sh somatic copy_numbers.control hom_snps.format het_snps.format hg19.fa
./haplotype_graph/haplotype_graph.sh
It takes a few hours during five iterations and outputs SVs, copy numbers and a breakpoint graph at the haplotype level.
Running JaBbA for comparison study.
jba junctions.vcf coverage.txt -p 3.4 -q 0.75
Check performance for SV calls from InfoGenomeR.
Rscript SV_performance.R SVs true_SV_sets_somatic 0 0
precision: 0.9552239 recall: 0.8050314 fmeasure: 0.8737201
Check performance for SV calls from JaBbA.
Rscript SV_performance.R jabba.simple.vcf.format true_SV_sets_somatic 0 0
precision: 0.9444444 recall: 0.7484277 fmeasure: 0.8350877

Demo 2

Run scripts for breakpoint graph construction.
./breakpoint_graph/breakpoint_graph.sh total A549 LUSC 2 4 bicseq_norm bicseq_norm_germ null hg19 5 null null 0
./breakpoint_graph/breakpoint_graph_simplifying.sh total A549 LUSC 2 4 bicseq_norm bicseq_norm_germ null hg19 5 null null 0
./allele_graph/allele_graph.sh total null hom_snps.format het_snps.format hg19.fa
./haplotype_graph/haplotype_graph.sh
It takes a day during 25 iterations (maximum 24 threads and 256Gb memory).
Run the script for plotting the haplotype graph.
Rscript ACN_GRAPH.R 3,8,11,15,19

Enter a directory for a subset of chromosomes, run the Eulerian path finding script, and run the plotting script for the candidate karyotype.
cd euler.15.19
./DAG.sh F
Rscript KAR_GRAPH.R

wisekh6 / InfoGenomeR