This repository contains code to create a docker implementation of the ADTEx tool.
The Aberration Detection in Tumour EXome (ADTEx) program was developed and published by KC Amarasinge et al (see References below). It is a tool to detect copy number variation (CNV) in exome sample pairs, usually a tumor and control from one patient.
In a comparative analysis of exome CNV callers (A Alkodsi et al), ADTEx performed better than most other callers tested.
Inputs to the program are a tumor/control pair of BAM files and several bed format helper files (see below). Output is a file with chromosome segments that are scored for amplification or deletion.
To get per-gene output, these scores must be mapped to an annotation, for example using [this program] (https://github.com/Jeltje/cnvtogenes)
ADTEx consists of a set of R scripts run through a python wrapper. The code includes a Circular Binary Segmentation (CBS) step in which short segments are merged into larger ones if their scores are similar (defined as a number of standard deviations usually between 0.5 and 3.5). The DNAcopy module that provides this calculation requires log score input. However, the original ADTEx code does not generate log ratios, instead it supplies straight ratios to the CBS tool. This results in a somewhat fragmented output.
In this implementation the python wrapper has been rewritten, and a post processing CBS step added.
The latest ADTEX docker image can be downloaded directly from quay.io using
docker pull quay.io/repository/ucsc_cgl/adtex
Alternatively, you can build from the github repo:
git clone https://github.com/BD2KGenomics/dockstore_tool_adtex
cd adtex
docker build -t ucsc_cgl/adtex .
For details on running docker containers in general, see the excellent tutorial at https://docs.docker.com/
To see a usage statement, run
docker run ucsc_cgl/adtex
docker run --log-driver=none -v /path/to/input/files:/data ucsc_cgl/adtex --normal normal.bam --tumor tumor.bam --sampleid MyTumorSample --out myOutputDir --targetbed targets.bed --centromeres centromeres.bed
where
normal.bam and tumor.bam are BAM format files of exome reads aligned to the genome. ADTEx will run the program bedtools to create coverage files. You can also supply coverage files directly, provided they were produced by bedtools v1.17 or below.
sampleid is an identifier for the patient. This will be used in the output.
out is the output directory, will be created in /path/to/input/files if it doesn't exist. See under Output for details. A subfolder named /tmp will be created inside this directory for temporary files.
targetbed is a list of exome targets in bed format
centromeres.bed is a bed format file containing centromere locations. This list is used to remove centromeres from the CBS calls.
Centromeres for hg19 are provided ind the /data directory
You can find centromere locations for genomes via http://genome.ucsc.edu/cgi-bin/hgTables Using the following selections:
- group: Mapping and Sequencing
- track:gap
- filter - goes to new page, look for 'type does match' and type centromere, submit
- output format: bed Submit, on the next page just press Get Bed
Since docker can only write to directories that are mounted via the -v parameter, the output directory is created inside
the input directory (/path/to/input/files in the above example). Inside the directory are the following files:
cnv.result is the original ADTEx output. It lists scores for all exons tested (derived from the targetbed file)
and a segment start, end, and ratio score.
<sampleID>.cnv is post processed CBS output. Here, the exon scores from the cnv.result file have been converted to log2 before
running the CBS algorithm, and exons have been merged into segments. The num.mark column shows the number of
exons in the segment. Seg_mean is a score in log2-1 format, which means that negative scores indicate deletions
and positive scores amplifications.
To get amplified or deleted segments from this file, a threshold must be applied. This is often set to 0.25/-0.25,
and with a minimum number of 10 markers per segment.
ADTEx can also calculate ploidy and contamination, if given an input of B allele frequencies (BAF). BAF files can be extracted from VCF records such as those generated by MuTect. These somatic point mutation callers must be run on the same input bamfiles.
The B allele is the non-reference genome allele of a SNP. ADTEx selects SNPs with two alleles that are heterozygous in the control sample, meaning that about 50% of the reads map to one allele, and 50% to the other. By looking at the read ratio in the tumor sample, ADTEx can predict which allele is preferentially amplified or lost. For instance, when 33% of reads map to allele P and 67 represent allele Q, there has been a selective amplification of the Q segment. If 100% of reads show the P allele, then the Q allele was lost. If ratios fall between these numbers, contamination of the tumor sample with non-tumor cells can be calculated.
To create BAF from VCF, run vcfToBaf.py from the /src directory. Note: This script has only been tested on MuTect output
To run ADTEx with BAF input:
docker run --log-driver=none -v /path/to/input/files:/data ucsc_cgl/adtex --normal normal.bam --tumor tumor.bam --sampleid MyTumorSample --out myOutputDir --targetbed targets.bed --centromeres centromeres.bed --estimatePloidy --baf MySample.baf
The zygosity step is run after the regular CNV caller, so the same output files will be generated as described before.
In addition:
zygosity.res is a file with cnv and zygosity calls per input SNP. The last column contains a human readable call on the SNP: LOH (loss of an allele), HET(erozygous), or ASCNA (allele specific amplification)
contamination contains a single number <1, indicating a fraction of the sample. When studying tumor samples, this is the faction of non-tumor RNA present in the sample.
ploidy contains the inferred ploidy of the tumor sample (2, 3, or 4). ADTEx assumes ploidy is not higher than 4.
Amarasinghe KC, Li J, Halgamuge SK. Correction: CoNVEX: copy number variation estimation in exome sequencing data using HMM. BMC Bioinformatics 2013;14 Suppl. 2:S26