GeneTrail 3.2

Advanced high-throughput enrichment analysis

Integrative analysis of Wilms' tumors

The Wilms’ tumor or nephroblastoma is the most common malignant renal cancer in children, affecting one in 10,000 children [1]. Approximately 75% of the cases occur in children younger than five years old, with a peak incidence in two- to three- year-olds [2]. Until recently, this tumor was always considered lethal, but serves today as a prime example of a curable malignant disease with survival rates of 90% [3]. Current therapeutic approaches rely on the classification of the tumor stage and histological subtype [4]. Identifying and understanding the different subtypes is crucial for finding a proper therapy and to assess the degree of malignancy of the tumor and thereby its aggressive potential. In order to describe the histology of the analyzed tumors the SIOP-classification scheme [5] was used. This classification scheme is the standard for renal tumors of the childhood that underwent a preoperative chemotherapy.

Analysis

In this use case we compare enrichment results computed from mRNA expression profiles to ones computed from miRNA expression profiles to investigate the differences between two distinct Wilms' tumors subtypes (regressive and blastema). For this analysis we use the mRNA enrichment analysis described in the transcriptomics use case. For the miRNA enrichment we use the sorted miRNA list of the miRNomics use case and select the top 10 up- and down-regulated miRNAs. Then we used the miRTarBase luciferase and CLASH assignment to map the contained miRNAs to the dedicated target genes. For each resulting gene list, we computed an ORA enrichment using all human genes as reference set. In the following sections, we showcase the comparative enrichment feature of GeneTrail2 to compare the mRNA and the miRNA results.

Technical Background

We cannot compare mRNA and miRNA enrichment analyses directly. To overcome this problem we map miRNA to the dedicated target genes and perform an enrichment analysis based on them. It is important to note that there is at the moment an ongoing discussion about about enrichment analysis based on miRNA targets. Bleazard et al. [6] show that over-representation analysis based on miRNA targets is influenced by an underlying bias in the predicted gene targets of miRNAs. The produced results often reflect this bias and contain many false positive predictions. Godard et al. [7] also report this bias and propose to use an alternative approach.

Parameter

mRNA
  • Test set: Ordered list of expressed mRNAs
  • Identifier-level statistics: Shrinkage t-test
  • Set-level statistics: Gene set enrichment analysis (GSEA)
  • P-value adjustment method: Benjamini-Hochberg
miRNA
  • Test set: Unordered list of deregulated miRNAs mapped to miRNA targets
  • Reference set: All miRNA targets
  • Set-level statistics: Over-representation analysis (ORA)
  • P-value adjustment method: Benjamini-Hochberg

Step-by-step slideshow

The following slideshow depicts the different analysis steps of the GeneTrail2 workflow.

Results

Since most miRNAs are assumed to trigger translational repression [8], we have to consider this fact to analyze the computed results. The targets of down-regulated miRNAs are assumed to be enriched and the targets of up-regulated miRNAs are assumed to be depleted.

Top 10% down-regulated miRNAs

For the top down-regulated miRNAs we expect their targets to be significantly enriched in the same categories as the mRNA data. This is in fact the case for all pathways that are involved in apoptosis and immune system.

Generic placeholder image
Generic placeholder image
Generic placeholder image

Top 10% down-regulated miRNA

For the top up-regulated miRNAs, we see a similar effect. All pathways that are involved in the proliferation of cells are significantly up-regulated in the miRNA data and down-regulated in mRNA data.

Generic placeholder image
Generic placeholder image

Bibliography

  1. Maschietto, Mariana and Piccoli, Fabio S and Costa, Cecilia ML and Camargo, Luiz P and Neves, Jose I and Grundy, Paul E and Brentani, Helena and Soares, Fernando A and de Camargo, Beatriz and Carraro, Dirce M Gene expression analysis of blastemal component reveals genes associated with relapse mechanism in Wilms tumour European Journal of Cancer Elsevier
  2. Dangio, G. J. Wilms tumor status report, 1990 Journal of Clinical Oncology WB Saunders Co Independence Square West Curtis Center, STE 300, Philadelphia, PA 19106-3399
  3. Graf, Norbert and Tournade, Marie-France and de Kraker, Jan The Role of Preoperative Chemotherapy in the Management of Wilms Tumor: The SIOP Studies Urologic Clinics of North America Elsevier
  4. Green, Daniel M The treatment of stages I--IV favorable histology Wilms' tumor Journal of clinical oncology American Society of Clinical Oncology
  5. Vujanic, Gordan M and Sandstedt, Bengt and Harms, Dieter and Kelsey, Anna and Leuschner, Ivo and de Kraker, Jan Revised International Society of Paediatric Oncology (SIOP) working classification of renal tumors of childhood Medical and pediatric oncology Wiley Online Library
  6. Bleazard, Thomas and Lamb, Janine A and Griffiths-Jones, Sam Bias in microRNA functional enrichment analysis Bioinformatics Oxford Univ Press
  7. Godard, Patrice and van Eyll, Jonathan Pathway analysis from lists of microRNAs: common pitfalls and alternative strategy Nucleic acids research Oxford Univ Press
  8. Wilczynska, A and Bushell, M The complexity of miRNA-mediated repression Cell Death and Differentiation Nature Publishing Group