nf-core/tumourevo

Edit

Introduction

This document describes the output produced by the pipeline. All plots generated in each step are summarised into the final report. The directories listed below will be created in the results directory after the pipeline has finished. All paths are relative to the top-level results directory.

Pipeline overview

The pipeline is built using Nextflow and processes data using the following steps:

• Variant Annotation
• Formatter
• Lifter
• Catalogue Driver Annotation
• QC
• Subclonal Deconvolution
• Signature Deconvolution

Directory Structure

The default directory structure is as follows:

{outdir}
├── variant_annotation
|   └── vep
│       └── <sample>
├── driver_annotation
|   └── annotate_driver
│       └── <sample>
├── pipeline_info
├── formatter
│   ├── cna2cnaqc
│       └── <sample>
│   ├── cnaqc2tsv
│       └── <patient>
|   └── vcf2cnaqc
│       └── <sample>
├── lifter
│   ├── mpileup
│       └── <patient>
│   └── positions
│       └── <sample>
├── QC
│   ├── tinc
│       └── <sample>
│   ├── CNAqc
│       └── <sample>
│   └── join_CNAqc
│       └── <patient>
├── signature_deconvolution
|   ├── SigProfiler
│       └── <dataset>
|   └── SparseSignatures
│       └── <dataset>
└── subclonal_deconvolution
|   ├── mobster
│       └── <sample>
|   ├── viber
│       └── <patient>
|   ├── ctree
│       └── <patient>,<sample>
|   └── pyclonevi
│       └── <patient>
work/
.nextflow.log

Variant Annotation

This directory contains results from the variant annotation subworkflow. At the level of individual samples, genomic variants present in the input VCF files are annotated using VEP.

VEP

VEP (Variant Effect Predictor) is a Ensembl tool that determines the effect of all variants (SNPs, insertions, deletions, CNVs or structural variants) on genes, transcripts, and protein sequence. This step starts from VCF files.

Formatter

The Formatter subworkflow is used to convert file to other formats and to standardize the output files resulting from different mutation (Mutect2, Strelka) and cna callers (ASCAT,Sequenza).

vcf2cnaqc

This parser is designed to process VCF files generated by various variant calling tools and to convert them into a unified RDS file format.

cna2cnaqc

This parser is designed to standardize copy number calls and purity estimates from various callers into a unified format.

cnaqc2tsv

This parser is designed to convert mutations data of joint CNAqc analysis from CNAqc format (RDS file) into a tabular format (TSV file). This step is mandatory for running python-based tools (e.g. PyClone-VI, SigProfiler) and it is mandatory if --tools contains either pyclone-vi or sigprofiler.

Lifter

The Lifter subworkflow is an optional step and it is run when single sample VCF file are provided. When multiple samples from the same patient are provided, the user can specify either a single joint VCF file, containing variant calls from all tumor samples of the patient (see joint variant calling), or individual sample specific VCF files. In the latter case, path to tumor BAM files must be provided in order to collect all mutations from the samples and perform pile-up of sample’s private mutations in all the other samples. Two intermediate steps, get_positions and mpileup, are performed to identify private mutations in all the samples and retrieve their variant allele frequency. Once private mutations are properly defined, they are merged back into the original VCF file during the join_positions step. The updated VCF file is then converted into a RDS object.

mpileup

At this stage, bcftools mpileup is run to retrieve frequency information of private mutations across all samples.

positions

This step allows to retrieve private and shared mutations across samples originated from the same patient. Previously retrieved mutations are joined with original mutations present in input VCF, which is in turn converted into an RDS object using vcfR.

Catalogue Driver Annotation

This directory contains results from the driver annotation subworkflow.

Tumour-type driver annotation

According to the specified tumour type, potential driver mutations are identified and annotated using IntOGen database. The user can also provide a custom table of driver genes.

QC

The QC subworkflows requires in input a segmentation file from allele-specific copy number callers (either Sequenza, ASCAT) and the joint VCF file. As a first step, the QC subworkflow provides an estimate of normal and tumour samples contamination in TINC step, in order to have a measure of experimental quality. Then,it first conducts a quality control on copy number and somatic mutation data for individual samples in CNAqc step, and subsequently summarize validated information at patient level in join_CNAqc step. The QC subworkflow is a crucial step of the pipeline as it ensures high confidence in identifying clonal and subclonal events while accounting for variations in tumor purity.

TINC

TINC is a package to calculate the contamination of tumor DNA in a matched normal sample. TINC provides estimates of the proportion of cancer cells containing the normal sample (TIN), and the proportion of cancer cells in the tumor sample (TIT).

CNAqc

CNAqc is a package that performs quality control of bulk cancer sequencing data for validating copy number segmentations against variant allele frequencies of somatic mutations.

join_CNAqc

This module creates a multi-CNAqc object for patient by summarizing the quality check performed at the single sample level. For more information about the structure of multi-CNAqc object see CNAqc documentation.

Signature Deconvolution

Mutational signatures are distinctive patterns of somatic mutations in cancer genomes that reveal the underlying mutational processes driving tumor evolution and progression. These signatures are identified by analyzing aggregated point-mutation counts from multiple samples. Validated mutations from the join_CNAqc step are converted into a joint TSV table (see cnaqc2tsv) and then input into the signature deconvolution subworkflow, which performs de novo extraction, inference, interpretation, or deconvolution of mutational counts.

The results of this step are collected in {pubslish_dir}/signature_deconvolution/. Two tools can be specified by using --tools parameter: SparseSignatures and SigProfiler.

SparseSignatures

SparseSignatures is an R-based computational framework that provides a set of functions to extract and visualize the mutational signatures that best explain the mutation counts of a large number of patients.

SigProfiler

SigProfiler is a python framework that allows de novo extraction of mutational signatures from data generated in a matrix format. The tool identifies the number of operative mutational signatures, their activities in each sample, and the probability for each signature to cause a specific mutation type in a cancer sample. The tool makes use of SigProfilerMatrixGenerator and SigProfilerExtractor, seamlessly integrating with other SigProfiler tools.

Subclonal Deconvolution

The subclonal deconvolution subworkflow requires in input a joint mCNAqc object resulting from the join_CNAqc step. The subworkflow will perform multi-sample deconvolution if more than one sample for each patient is present.

The results of subclonal decovnultion step are collected in {outdir}/subclonal_deconvolution/ directory.

MOBSTER

MOBSTER is a package that models mutant allelic frequencies and copy-number status by integrating evolutionary theory and Bayesian proabilistic modelling to identify clusters of variants with similar cellular proportions. Futhermore, MOBSTER models the dynamics of passenger mutations via a Pareto distribution giving rise to the so called neutral tail.

PyClone-VI

PyClone-VI is a computationally efficient Bayesian statistical method for inferring the clonal population structure of cancers, by considering allele fractions and coincident copy number variation using a variational inference approach. It works for patients with both single and multiple samples.

VIBER

VIBER is an R package that implements a variational Bayesian model to fit multi-variate Binomial mixtures. In the context of subclonal deconvolution VIBER models read counts that are associated with the most represented karyotype. It works for patients with both single and multiple samples.

ctree

Subclonal deconvolution results are used to build clone tree from both single samples and multple samples using ctree. ctree is a R-based package which implements basic functions to create, manipulate and visualize clone trees by modelling Cancer Cell Fractions (CCF) clusters. Annotated driver genes must be provided in the input data.

NB: When --tools pyclone-vi is used, the output of PyClone-VI subclonal deconvolution is preprocessed prior to clone tree inference. Since ctree requires labeling one of the clusters as “clonal,” the one with the highest CCF across all samples is choosen.

VIBER and MOBSTER fits are already compatible for ctree analysis.

Output directory: {outdir}/subclonal_deconvolution/ctree/<patient>/

• ctree_<tool>.rds
  • RDS file containing inferred clone tree
• ctree_<tool>_plots.rds
  • RDS file for single sample clone tree plot
• ctree_input_pyclonevi.csv
  • CSV file required for clone tree inference from pyclone

Pipeline information

Nextflow provides excellent functionality for generating various reports relevant to the running and execution of the pipeline. This will allow you to troubleshoot errors with the running of the pipeline, and also provide you with other information such as launch commands, run times and resource usage.

Reference files

Different tools of the pipeline generate references files. Once reference file for VEP and SigProfiler are not provided they are stored in the tool-specific folder.

VEP

When VEP cache is not specified, the desired VEP cache is downladed in {outdir}/references/VEP/vep_cache/homo_sapiens/{VEP_version}_{ref_genome}.

SigProfiler

Reference genome for SigProfiler is store in the following folder: