umccr/wtsreport repository overview

⁠RNA-seq report

RNA-seq report workflow is designed to post-process, summarise and visualise an output from bcbio-nextgen⁠ RNA-seq pipeline⁠. Its main application is to complement genome-based findings from umccrise⁠ pipeline and to provide additional evidence for detected alterations.

⁠Workflow

The pipeline consist of four main components illustrated and breifly described below.

1. Process the patient sample WTS from bcbio-nextgen⁠ RNA-seq pipeline⁠ including per-gene read counts⁠ and gene fusions⁠, and add expression data from reference cohorts⁠ to get an idea about expression levels of genes of interest in other cancer patient cohorts⁠. The gene fusion⁠ candidates are re-quantified and the read counts are normalised, transformed⁠ and converted⁠ into a scale that allows to present the sample's expression measurements in the context of the reference cohorts⁠.

2. Feed in genome-based findings from whole-genome sequencing (WGS) data to focus on genes of interest and provide additional evidence for dysregulation of mutated genes, or genes located within detected structural variants (SVs) or copy-number (CN) altered regions. The RNA-seq report pipeline is designed to be compatible with WGS patient report based on umccrise⁠ pipeline output.

3. Collate results with knowledge derived from in-house resources and public databases to provide additional source of evidence for clinical significance of altered genes, e.g. to flag variants with clinical significance or potential druggable targets.

4. The final product is a html-based interactive report with searchable tables and plots presenting expression levels of the genes of interest genes. The report consist of several sections described in report_structure.md⁠.

⁠Reference data

The reference expression data is availbale for 33 cancer types and were derived from external⁠ (TCGA⁠) and internal⁠ (UMCCR⁠) resources.

⁠External reference cohorts

In order to explore expression changes in queried sample we have built a high-quality pancreatic cancer reference cohort.

Depending on the tissue from which the patient's sample was taken, one of 33 cancer datasets from TCGA⁠ can be used as a reference cohort for comparing expression changes in genes of interest in investigated sample. The available cancer types are listed in TCGA projects summary table⁠. These datasets have been processed using methods described in TCGA-data-harmonization⁠ repository. The dataset of interest can be specified by using one of the TCGA⁠ project IDs (Project column) for the --dataset argument.

⁠Note

Each dataset was cleaned based on the quality metrics provided in the Merged Sample Quality Annotations file merged_sample_quality_annotations.tsv⁠ from TCGA PanCanAtlas initiative webpage⁠ (see TCGA-data-harmonization⁠ repository for more details, including sample inclusion criteria).

⁠Internal reference cohort

The publically available TCGA datasets are expected to demonstrate prominent batch effects⁠ when compared to the in-house WTS data due to differences in applied experimental procedures and analytical pipelines. Moreover, TCGA data may include samples from tissue material of lower quality and cellularity compared to samples processed using local protocols. To address these issues, we have built a high-quality internal reference cohort processed using the same pipelines as input data.

This internal reference set of 40 pancreatic cancer samples is based on WTS data generated at UMCCR⁠ and processed with bcbio-nextgen⁠ RNA-seq pipeline⁠ to minimise potential batch effects between investigated samples and the reference cohort and to make sure the data are comparable. The internal reference cohort assembly is summarised in Pancreatic-data-harmonization⁠ repository.

⁠Note

The are two rationales for using the internal reference cohort:

1. In case of pancreatic cancer samples this cohort serves as a reference point for comparing per-gene expression levels observed in the investigated single-subject data and data from other pancreatic cancer patients.

2. In case of samples from any cancer type the data from the internal reference cohort is used in batch effects correction procedure performed to minimise technical-related variation in the data.

⁠Input data

The pipeline accepts WTS⁠ data processed by bcbio-nextgen⁠ RNA-seq pipeline⁠. Additionally, the WTS data can be integrated with WGS⁠-based data processed using umccrise⁠ pipeline. In the latter case, the genome-based findings from corresponding sample are incorporated into the report and are used as a primary source for expression profiles prioritisation.

⁠WTS

The only required WTS input data are read counts provided in quantification file from kallisto⁠. The per-transcript abundances are reported in estimated counts (est_counts) and in Transcripts Per Million⁠ (tpm), which are then converted to per-gene estimates. Additionally, a list of fusion genes⁠ detected by pizzly⁠ can be provided.

Table below lists all input data accepted in the pipeline: