DECoN – detecting tricky gene mutations

DECoNThe TGMI are pleased to announce a new software tool, DECoN, that we have made available. We have also published a paper detailing the development, evaluation and implementation of DECoN in Wellcome Open Research today.

DECoN is an acronym for Detection of Exon Copy Number variants. DECoN allows a tricky class of gene mutation, known as exon CNVs, to be automatically detected through analysis of gene panel sequencing data. Typically, most labs use an entirely separate test to detect this type of mutation. Inevitably this adds time and cost, and such tests are not available for all genes.


Genes can be mutated in many different ways

The most common way for genes to be mutated is through one letter of DNA being swapped for a different letter. These are called ‘substitutions’. Deletions or insertions of a few letters of DNA also occur not uncommonly. These are called ‘indels’.

We all have hundreds of substitutions and indels and most are innocuous. But if they occur at critical points in critical genes these mutations cause genetic conditions. Together they are the predominant mutation types that cause genetic disease.

There are, though, many other mutation types and ways in which gene function can be altered. Although rarer, these are still very important, not least because they can also lead to genetic conditions.


Exon CNVs are an important cause of genetic conditions

Exon CNVs are an example of these less common, but important, mutation types.

Genes are made up of blocks, called exons, as I described in a previous blog. Sometimes, one or more whole exons are deleted or duplicated. Several different terms are used to describe this mutation type. We use the term ‘exon copy number variant’, and the short term ‘exon CNV’.

Exon CNVs are a key cause of genetic conditions. For example, 10% of BRCA1 cancer causing mutations are exon CNVs

Exon CNVs typically lead to the gene code becoming scrambled and they often stop the gene working at all. We have two copies of each gene so this doesn’t necessarily cause any major problems.

But exon CNVs that occur in human disease genes are an important cause of genetic conditions. For example, about 10 percent of BRCA1 cancer causing mutations are exon CNVs.


Exon CNVs are tricky to detect

DNA sequence can easily and accurately be analysed to detect substitutions and small indels. Exon CNVs have been more challenging to detect, for many different reasons, which we discuss in the DECoN paper.

In the research setting, this has led to exon CNVs typically being ignored. In the clinical setting, this is simply not acceptable for genes like BRCA1. As a result, many different methods to detect exon CNVs have been developed and deployed.

In my lab, we have mostly used MLPA to identify exon CNVs. Whilst accurate, it is costly, only suitable for small numbers of samples and is a separate test that needs to be done in addition to our sequencing test.


Accurate detection of exon CNVs in sequence data is now achievable

Rapid, accurate, high-throughput detection of exon CNVs has been a headache and a bottleneck in our desire to deliver fast, affordable, BRCA gene testing.  Ideally, we wanted to be able to use the DNA sequence data being generated to detect substitutions and indels to also detect exon CNVs. This would allow us to have an integrated, single pipeline for all three mutation types.

Fortunately for us, Vincent Plagnol and his collaborators had developed an algorithm called ExomeDepth that can do this. Several TGMI investigators have been exploring the use of ExomeDepth to detect exon CNVs in gene panel data in the clinical setting. It works really well. Not yet perfectly, but very very well.


DECoN is a fast, accurate, exon CNV detection tool

My own group, together with Gerton Lunter’s group at The Wellcome Trust Centre for Human Genetics have focused on making a free, easy-to-use tool that optimises ExomeDepth for the clinical setting. DECoN uses the sequence data generated from the cancer gene panel that we run on 96 samples each week to generate exon CNV calls. This is fully automated and only adds 30 mins to the end of the analysis step for all 96 samples. We do confirm exon CNVs detected by DECoN by MLPA.

DECoN is very good, though there is also room for improvement, as we discuss in the paper. After undertaking 1000s of simulated and experimental evaluations we have implemented DECoN in our clinical testing lab, and its use has been approved by our accreditation body, UKAS.

DECoN has been central to our ability to roll out fast, affordable, routine cancer gene testing.


DECoN is freely available

One of the TGMI’s aims is to develop and validate flexible processes that maximise the research and clinical utilities of genetic testing. DECoN is an example of an output of this aim. To facilitate its clinical usefulness we have made an easy-to-install, packaged version available at together with comprehensive documentation.

We have also made DECoN available on GitHub for the research user.

We hope DECoN will be useful to many, and would be happy to receive any comments or feedback.



Image: DECoN picture showing a BRCA1 exon CNV. BRCA1 contains 24 exons. Exons 8 through 13 are deleted (in red) in this sample. See our paper for more details.

Download the press release.