Yvan Saeys

Robust Feature Selection Using Ensemble Feature Selection Techniques

Robustness or stability of feature selection techniques is a topic of recent interest, and is an important issue when selected feature subsets are subsequently analysed by domain experts to gain more insight into the problem modelled. In this work, we investigate the use of ensemble feature selection techniques, where multiple feature selection methods are combined to yield more robust results. We show that these techniques show great promise for high-dimensional domains with small sample sizes, and provide more robust feature subsets than a single feature selection technique. In addition, we also investigate the effect of ensemble feature selection techniques on classification performance, giving rise to a new model selection strategy.

Robust biomarker identification for cancer diagnosis with ensemble feature selection methods

MOTIVATION Biomarker discovery is an important topic in biomedical applications of computational biology, including applications such as gene and SNP selection from high-dimensional data. Surprisingly, the stability with respect to sampling variation or robustness of such selection processes has received attention only recently. However, robustness of biomarkers is an important issue, as it may greatly influence subsequent biological validations. In addition, a more robust set of markers may strengthen the confidence of an expert in the results of a selection method. RESULTS Our first contribution is a general framework for the analysis of the robustness of a biomarker selection algorithm. Secondly, we conducted a large-scale analysis of the recently introduced concept of ensemble feature selection, where multiple feature selections are combined in order to increase the robustness of the final set of selected features. We focus on selection methods that are embedded in the estimation of support vector machines (SVMs). SVMs are powerful classification models that have shown state-of-the-art performance on several diagnosis and prognosis tasks on biological data. Their feature selection extensions also offered good results for gene selection tasks. We show that the robustness of SVMs for biomarker discovery can be substantially increased by using ensemble feature selection techniques, while at the same time improving upon classification performances. The proposed methodology is evaluated on four microarray datasets showing increases of up to almost 30% in robustness of the selected biomarkers, along with an improvement of approximately 15% in classification performance. The stability improvement with ensemble methods is particularly noticeable for small signature sizes (a few tens of genes), which is most relevant for the design of a diagnosis or prognosis model from a gene signature. SUPPLEMENTARY INFORMATION Supplementary data are available at Bioinformatics online.

The function of Fcγ receptors in dendritic cells and macrophages

Dendritic cells (DCs) and macrophages use various receptors to recognize foreign antigens and to receive feedback control from adaptive immune cells. Although it was long believed that all immunoglobulin Fc receptors are universally expressed by phagocytes, recent findings indicate that only monocyte-derived DCs and macrophages express high levels of activating Fc receptors for IgG (FcγRs), whereas conventional and plasmacytoid DCs express the inhibitory FcγR. In this Review, we discuss how the uptake, processing and presentation of antigens by DCs and macrophages is influenced by FcγR recognition of immunoglobulins and immune complexes in the steady state and during inflammation.

Bone marrow-derived monocytes give rise to self-renewing and fully differentiated Kupffer cells

Self-renewing tissue-resident macrophages are thought to be exclusively derived from embryonic progenitors. However, whether circulating monocytes can also give rise to such macrophages has not been formally investigated. Here we use a new model of diphtheria toxin-mediated depletion of liver-resident Kupffer cells to generate niche availability and show that circulating monocytes engraft in the liver, gradually adopt the transcriptional profile of their depleted counterparts and become long-lived self-renewing cells. Underlining the physiological relevance of our findings, circulating monocytes also contribute to the expanding pool of macrophages in the liver shortly after birth, when macrophage niches become available during normal organ growth. Thus, like embryonic precursors, monocytes can and do give rise to self-renewing tissue-resident macrophages if the niche is available to them.

Unsupervised High-Dimensional Analysis Aligns Dendritic Cells across Tissues and Species.

Summary Dendritic cells (DCs) are professional antigen-presenting cells that hold great therapeutic potential. Multiple DC subsets have been described, and it remains challenging to align them across tissues and species to analyze their function in the absence of macrophage contamination. Here, we provide and validate a universal toolbox for the automated identification of DCs through unsupervised analysis of conventional flow cytometry and mass cytometry data obtained from multiple mouse, macaque, and human tissues. The use of a minimal set of lineage-imprinted markers was sufficient to subdivide DCs into conventional type 1 (cDC1s), conventional type 2 (cDC2s), and plasmacytoid DCs (pDCs) across tissues and species. This way, a large number of additional markers can still be used to further characterize the heterogeneity of DCs across tissues and during inflammation. This framework represents the way forward to a universal, high-throughput, and standardized analysis of DC populations from mutant mice and human patients.

Yolk Sac Macrophages, Fetal Liver, and Adult Monocytes Can Colonize an Empty Niche and Develop into Functional Tissue-Resident Macrophages

Tissue-resident macrophages can derive from yolk sac macrophages (YS-Macs), fetal liver monocytes (FL-MOs), or adult bone-marrow monocytes (BM-MOs). The relative capacity of these precursors to colonize a niche, self-maintain, and perform tissue-specific functions is unknown. We simultaneously transferred traceable YS-Macs, FL-MOs, and BM-MOs into the empty alveolar macrophage (AM) niche of neonatal Csf2rb(-/-) mice. All subsets produced AMs, but in competition preferential outgrowth of FL-MOs was observed, correlating with their superior granulocyte macrophage-colony stimulating factor (GM-CSF) reactivity and proliferation capacity. When transferred separately, however, all precursors efficiently colonized the alveolar niche and generated AMs that were transcriptionally almost identical, self-maintained, and durably prevented alveolar proteinosis. Mature liver, peritoneal, or colon macrophages could not efficiently colonize the empty AM niche, whereas mature AMs could. Thus, precursor origin does not affect the development of functional self-maintaining tissue-resident macrophages and the plasticity of the mononuclear phagocyte system is largest at the precursor stage.

The unfolded-protein-response sensor IRE-1α regulates the function of CD8α + dendritic cells

The role of the unfolded protein response (UPR) and endoplasmic reticulum (ER) stress in homeostasis of the immune system is incompletely understood. Here we found that dendritic cells (DCs) constitutively activated the UPR sensor IRE-1α and its target, the transcription factor XBP-1, in the absence of ER stress. Loss of XBP-1 in CD11c+ cells led to defects in phenotype, ER homeostasis and antigen presentation by CD8α+ conventional DCs, yet the closely related CD11b+ DCs were unaffected. Whereas the dysregulated ER in XBP-1-deficient DCs resulted from loss of XBP-1 transcriptional activity, the phenotypic and functional defects resulted from regulated IRE-1α-dependent degradation (RIDD) of mRNAs, including those encoding CD18 integrins and components of the major histocompatibility complex (MHC) class I machinery. Thus, a precisely regulated feedback circuit involving IRE-1α and XBP-1 controls the homeostasis of CD8α+ conventional DCs.

Large-scale structural analysis of the core promoter in mammalian and plant genomes

DNA encodes at least two independent levels of functional information. The first level is for encoding proteins and sequence targets for DNA-binding factors, while the second one is contained in the physical and structural properties of the DNA molecule itself. Although the physical and structural properties are ultimately determined by the nucleotide sequence itself, the cell exploits these properties in a way in which the sequence itself plays no role other than to support or facilitate certain spatial structures. In this work, we focus on these structural properties, comparing them between different organisms and assessing their ability to describe the core promoter. We prove the existence of distinct types of core promoters, based on a clustering of their structural profiles. These results indicate that the structural profiles are much conserved within plants (Arabidopsis and rice) and animals (human and mouse), but differ considerably between plants and animals. Furthermore, we demonstrate that these structural profiles can be an alternative way of describing the core promoter, in addition to more classical motif or IUPAC-based approaches. Using the structural profiles as discriminatory elements to separate promoter regions from non-promoter regions, reliable models can be built to identify core-promoter regions using a strictly computational approach.

GenomeView: a next-generation genome browser

Due to ongoing advances in sequencing technologies, billions of nucleotide sequences are now produced on a daily basis. A major challenge is to visualize these data for further downstream analysis. To this end, we present GenomeView, a stand-alone genome browser specifically designed to visualize and manipulate a multitude of genomics data. GenomeView enables users to dynamically browse high volumes of aligned short-read data, with dynamic navigation and semantic zooming, from the whole genome level to the single nucleotide. At the same time, the tool enables visualization of whole genome alignments of dozens of genomes relative to a reference sequence. GenomeView is unique in its capability to interactively handle huge data sets consisting of tens of aligned genomes, thousands of annotation features and millions of mapped short reads both as viewer and editor. GenomeView is freely available as an open source software package.

Computational flow cytometry: helping to make sense of high-dimensional immunology data

Recent advances in flow cytometry allow scientists to measure an increasing number of parameters per cell, generating huge and high-dimensional datasets. To analyse, visualize and interpret these data, newly available computational techniques should be adopted, evaluated and improved upon by the immunological community. Computational flow cytometry is emerging as an important new field at the intersection of immunology and computational biology; it allows new biological knowledge to be extracted from high-throughput single-cell data. This Review provides non-experts with a broad and practical overview of the many recent developments in computational flow cytometry.