Privacy preserving validation for multiomic prediction models.

Reproducibility of results obtained using ribonucleic acid (RNA) data across labs remains a major hurdle in cancer research. Often, molecular predictors trained on one dataset cannot be applied to another due to differences in RNA library preparation and quantification, which inhibits the validation of predictors across labs. While current RNA correction algorithms reduce these differences, they require simultaneous access to patient-level data from all datasets, which necessitates the sharing of training data for predictors when sharing predictors. Here, we describe SpinAdapt, an unsupervised RNA correction algorithm that enables the transfer of molecular models without requiring access to patient-level data. It computes data corrections only via aggregate statistics of each dataset, thereby maintaining patient data privacy. Despite an inherent trade-off between privacy and performance, SpinAdapt outperforms current correction methods, like Seurat and ComBat, on publicly available cancer studies, including TCGA and ICGC. Furthermore, SpinAdapt can correct new samples, thereby enabling unbiased evaluation on validation cohorts. We expect this novel correction paradigm to enhance research reproducibility and to preserve patient privacy.

An integrated model for detecting significant chromatin interactions from high-resolution Hi-C data.

Here we present HiC-DC, a principled method to estimate the statistical significance (P values) of chromatin interactions from Hi-C experiments. HiC-DC uses hurdle negative binomial regression account for systematic sources of variation in Hi-C read counts-for example, distance-dependent random polymer ligation and GC content and mappability bias-and model zero inflation and overdispersion. Applied to high-resolution Hi-C data in a lymphoblastoid cell line, HiC-DC detects significant interactions at the sub-topologically associating domain level, identifying potential structural and regulatory interactions supported by CTCF binding sites, DNase accessibility, and/or active histone marks. CTCF-associated interactions are most strongly enriched in the middle genomic distance range (∼700 kb-1.5 Mb), while interactions involving actively marked DNase accessible elements are enriched both at short (<500 kb) and longer (>1.5 Mb) genomic distances. There is a striking enrichment of longer-range interactions connecting replication-dependent histone genes on chromosome 6, potentially representing the chromatin architecture at the histone locus body.

Practical Analysis of Genome Contact Interaction Experiments.

The three dimensional (3D) architecture of chromosomes is not random but instead tightly organized due to chromatin folding and chromatin interactions between genomically distant loci. By bringing genomically distant functional elements such as enhancers and promoters into close proximity, these interactions play a key role in regulating gene expression. Some of these interactions are dynamic, that is, they differ between cell types, conditions and can be induced by specific stimuli or differentiation events. Other interactions are more structural and stable, that is they are constitutionally present across several cell types. Genome contact interactions can occur via recruitment and physical interaction between chromatin-binding proteins and correlate with epigenetic marks such as histone modifications. Absence of a contact can occur due to presence of insulators, that is, chromatin-bound complexes that physically separate genomic loci. Understanding which contacts occur or do not occur in a given cell type is important since it can help explain how genes are regulated and which functional elements are involved in such regulation. The analysis of genome contact interactions has been greatly facilitated by the relatively recent development of chromosome conformation capture (3C). In an even more recent development, 3C was combined with next generation sequencing and led to Hi-C, a technique that in theory queries all possible pairwise interactions both within the same chromosome (intra) and between chromosomes (inter). Hi-C has now been used to study genome contact interactions in several human and mouse cell types as well as in animal models such as Drosophila and yeast. While it is fair to say that Hi-C has revolutionized the study of chromatin interactions, the computational analysis of Hi-C data is extremely challenging due to the presence of biases, artifacts, random polymer ligation and the huge number of potential pairwise interactions. In this chapter, we outline a strategy for analysis of genome contact experiments based on Hi-C using R and Bioconductor.

Characterization of a set of tumor suppressor microRNAs in T cell acute lymphoblastic leukemia.

The posttranscriptional control of gene expression by microRNAs (miRNAs) is highly redundant, and compensatory effects limit the consequences of the inactivation of individual miRNAs. This implies that only a few miRNAs can function as effective tumor suppressors. It is also the basis of our strategy to define functionally relevant miRNA target genes that are not under redundant control by other miRNAs. We identified a functionally interconnected group of miRNAs that exhibited a reduced abundance in leukemia cells from patients with T cell acute lymphoblastic leukemia (T-ALL). To pinpoint relevant target genes, we applied a machine learning approach to eliminate genes that were subject to redundant miRNA-mediated control and to identify those genes that were exclusively targeted by tumor-suppressive miRNAs. This strategy revealed the convergence of a small group of tumor suppressor miRNAs on the Myb oncogene, as well as their effects on HBP1, which encodes a transcription factor. The expression of both genes was increased in T-ALL patient samples, and each gene promoted the progression of T-ALL in mice. Hence, our systematic analysis of tumor suppressor miRNA action identified a widespread mechanism of oncogene activation in T-ALL.