CANCERSIGN: a user-friendly and robust tool for identification and classification of mutational signatures and patterns in cancer genomes

Abstract

Analysis of cancer mutational signatures have been instrumental in identification of responsible endogenous and exogenous molecular processes in cancer. The quantitative approach used to deconvolute mutational signatures is becoming an integral part of cancer research. Therefore, development of a stand-alone tool with a user-friendly interface for analysis of cancer mutational signatures is necessary. In this manuscript we introduce CANCERSIGN, which enables users to identify 3-mer and 5-mer mutational signatures within whole genome, whole exome or pooled samples. Additionally, this tool enables users to perform clustering on tumor samples based on the proportion of mutational signatures in each sample. Using CANCERSIGN, we analysed all the whole genome somatic mutation datasets profiled by the International Cancer Genome Consortium (ICGC) and identified a number of novel signatures. By examining signatures found in exonic and non-exonic regions of the genome using WGS and comparing this to signatures found in WES data we observe that WGS can identify additional non-exonic signatures that are enriched in the non-coding regions of the genome while the deeper sequencing of WES may help identify weak signatures that are otherwise missed in shallower WGS data. Aberrant somatic changes in DNA resulting from endogenous sources (e.g. APOBEC-induced mutagenesis and DNA repair defects) and exogenous factors (e.g. tobacco smoking and UV radiation) are the hallmark of cancer. These alternations in DNA may have different forms, ranging from gross chromosomal rearrangements to single base substitutions 1. The whole genome sequencing of tumor cells has shown that the number of mutations varies from less than one hundred per genome to hundreds of thousands depending on the cancer type and patient. Moreover, the type of mutation and sequence context of many cancer mutations are not random. For instance, C-toT mutation within the CG (a.k.a. CpG) dinucleotide is a prevalent mutation in cancer and as its abundance is proportional to the age of patient it is referred to as an "aging" signature 2. Many cancers also have a large number of C-toT and C-to-G mutations within TCA and TCT trinucleotides 3. These mutations are attributed to the aberrant changes in the level and activity of APOBEC enzymes. The mutational landscape of each cancer genome is thus a cumulative result of multiple mutational signatures, each caused by a unique process such as methylation, APOBEC mediated changes, etc. 1. Typically, signatures of mutational processes are determined by considering the trinucleotide context of single base substitutions. If all mutations are presented based on changes in the same DNA strand, there are 96 possible

Integrative analysis of mutated genes and mutational processes reveals novel mutational biomarkers in colorectal cancer

“…These CRC mutational signatures are associated to age of patients (signature 1), defective DNA mismatch repair (signature 6), and altered activity of the error-prone polymerase POLE (signature 10) [ 56 ]. To find which mutational processes are active in the CRC subtypes, we used the CANCERSIGN tool [ 57 ] to identify mutational signatures ( see the method section ). Using whole genome sequencing data from ICGC CRC samples, we identify seven signatures overall for the CRC patients (Fig.…”

“…We also use CANCERSIGN [ 57 ] to identify signatures that are represented in our CRC samples. CANCERSIGN uses a non-negative matrix factorization (NMF) algorithm to identify optimal number of signatures.…”

Whole-Genome Analysis of De Novo Somatic Point Mutations Reveals Novel Mutational Biomarkers in Pancreatic Cancer

“…We used the CANCERSIGN package in R [27] to calculate the mutational signatures of pancreatic cancer, and the level of exposures of each sample to each signature. This tool implements the Non-negative Matrix Factorization (NMF) method to find patterns of 3-nucleotide motifs among samples.…”

“…As it was discussed in [43], different signatures can be interpreted as different molecular mechanisms of mutations. Here, we used CANCERSIGN tool [27] to identify mutational signatures in the identified PC subtypes. Patterns of extracted mutational signatures are provided in Figure S7 (considering Alexnadrov et al signatures profile, we excluded unknown and artifact signatures from our analysis) [28].…”

Deciphering Genomic Heterogeneity and the Internal Composition of Tumour Activities through a Hierarchical Factorisation Model

“…The obtained model is usually a low-rank representation, describing in some cases NMF as a dimensionality reduction technique. NMF has been particularly useful in computational biology, providing relevant contributions in the field of clustering [10], protein-protein interaction analysis [11], deconvolution of mutational patterns [12,13], as well as inferring the proportion of different cell types in tissue samples [14,15], among others.…”