Marcel J. T. Reinders

Unifying user-based and item-based collaborative filtering approaches by similarity fusion

Memory-based methods for collaborative filtering predict new ratings by averaging (weighted) ratings between, respectively, pairs of similar users or items. In practice, a large number of ratings from similar users or similar items are not available, due to the sparsity inherent to rating data. Consequently, prediction quality can be poor. This paper re-formulates the memory-based collaborative filtering problem in a generative probabilistic framework, treating individual user-item ratings as predictors of missing ratings. The final rating is estimated by fusing predictions from three sources: predictions based on ratings of the same item by other users, predictions based on different item ratings made by the same user, and, third, ratings predicted based on data from other but similar users rating other but similar items. Existing user-based and item-based approaches correspond to the two simple cases of our framework. The complete model is however more robust to data sparsity, because the different types of ratings are used in concert, while additional ratings from similar users towards similar items are employed as a background model to smooth the predictions. Experiments demonstrate that the proposed methods are indeed more robust against data sparsity and give better recommendations.

Molecular Maps of the Reorganization of Genome-Nuclear Lamina Interactions during Differentiation

The three-dimensional organization of chromosomes within the nucleus and its dynamics during differentiation are largely unknown. To visualize this process in molecular detail, we generated high-resolution maps of genome-nuclear lamina interactions during subsequent differentiation of mouse embryonic stem cells via lineage-committed neural precursor cells into terminally differentiated astrocytes. This reveals that a basal chromosome architecture present in embryonic stem cells is cumulatively altered at hundreds of sites during lineage commitment and subsequent terminal differentiation. This remodeling involves both individual transcription units and multigene regions and affects many genes that determine cellular identity. Often, genes that move away from the lamina are concomitantly activated; many others, however, remain inactive yet become unlocked for activation in a next differentiation step. These results suggest that lamina-genome interactions are widely involved in the control of gene expression programs during lineage commitment and terminal differentiation.

An expression profile for diagnosis of lymph node metastases from primary head and neck squamous cell carcinomas

Metastasis is the process by which cancers spread to distinct sites in the body. It is the principal cause of death in individuals suffering from cancer. For some types of cancer, early detection of metastasis at lymph nodes close to the site of the primary tumor is pivotal for appropriate treatment. Because it can be difficult to detect lymph node metastases reliably, many individuals currently receive inappropriate treatment. We show here that DNA microarray gene-expression profiling can detect lymph node metastases for primary head and neck squamous cell carcinomas that arise in the oral cavity and oropharynx. The predictor, established with an 82-tumor training set, outperforms current clinical diagnosis when independently validated. The 102 predictor genes offer unique insights into the processes underlying metastasis. The results show that the metastatic state can be deciphered from the primary tumor gene-expression pattern and that treatment can be substantially improved.

Resolving motion correspondence for densely moving points

Studies the motion correspondence problem for which a diversity of qualitative and statistical solutions exist. We concentrate on qualitative modeling, especially in situations where assignment conflicts arise either because multiple features compete for one detected point or because multiple detected points fit a single feature point. We leave out the possibility of point track initiation and termination because that principally conflicts with allowing for temporary point occlusion. We introduce individual, combined, and global motion models and fit existing qualitative solutions in this framework. Additionally, we present a tracking algorithm that satisfies these-possibly constrained-models in a greedy matching sense, including an effective way to handle detection errors and occlusion. The performance evaluation shows that the proposed algorithm outperforms existing greedy matching algorithms. Finally, we describe an extension to the tracker that enables automatic initialization of the point tracks. Several experiments show that the extended algorithm is efficient, hardly sensitive to its few parameters, and qualitatively better than other algorithms, including the presumed optimal statistical multiple hypothesis tracker.

New insights on human T cell development by quantitative T cell receptor gene rearrangement studies and gene expression profiling

To gain more insight into initiation and regulation of T cell receptor (TCR) gene rearrangement during human T cell development, we analyzed TCR gene rearrangements by quantitative PCR analysis in nine consecutive T cell developmental stages, including CD34+ lin− cord blood cells as a reference. The same stages were used for gene expression profiling using DNA microarrays. We show that TCR loci rearrange in a highly ordered way (TCRD-TCRG-TCRB-TCRA) and that the initiating Dδ2-Dδ3 rearrangement occurs at the most immature CD34+CD38−CD1a− stage. TCRB rearrangement starts at the CD34+CD38+CD1a− stage and complete in-frame TCRB rearrangements were first detected in the immature single positive stage. TCRB rearrangement data together with the PTCRA (pTα) expression pattern show that human TCRβ-selection occurs at the CD34+CD38+CD1a+ stage. By combining the TCR rearrangement data with gene expression data, we identified candidate factors for the initiation/regulation of TCR recombination. Our data demonstrate that a number of key events occur earlier than assumed previously; therefore, human T cell development is much more similar to murine T cell development than reported before.

An algorithm-based topographical biomaterials library to instruct cell fate

It is increasingly recognized that material surface topography is able to evoke specific cellular responses, endowing materials with instructive properties that were formerly reserved for growth factors. This opens the window to improve upon, in a cost-effective manner, biological performance of any surface used in the human body. Unfortunately, the interplay between surface topographies and cell behavior is complex and still incompletely understood. Rational approaches to search for bioactive surfaces will therefore omit previously unperceived interactions. Hence, in the present study, we use mathematical algorithms to design nonbiased, random surface features and produce chips of poly(lactic acid) with 2,176 different topographies. With human mesenchymal stromal cells (hMSCs) grown on the chips and using high-content imaging, we reveal unique, formerly unknown, surface topographies that are able to induce MSC proliferation or osteogenic differentiation. Moreover, we correlate parameters of the mathematical algorithms to cellular responses, which yield novel design criteria for these particular parameters. In conclusion, we demonstrate that randomized libraries of surface topographies can be broadly applied to unravel the interplay between cells and surface topography and to find improved material surfaces.

A maximum variance cluster algorithm

We present a partitional cluster algorithm that minimizes the sum-of-squared-error criterion while imposing a hard constraint on the cluster variance. Conceptually, hypothesized clusters act in parallel and cooperate with their neighboring clusters in order to minimize the criterion and to satisfy the variance constraint. In order to enable the demarcation of the cluster neighborhood without crucial parameters, we introduce the notion of foreign cluster samples. Finally, we demonstrate a new method for cluster tendency assessment based on varying the variance constraint parameter.

Large-Scale Mutagenesis in p19ARF- and p53-Deficient Mice Identifies Cancer Genes and Their Collaborative Networks

Summary p53 and p19ARF are tumor suppressors frequently mutated in human tumors. In a high-throughput screen in mice for mutations collaborating with either p53 or p19ARF deficiency, we identified 10,806 retroviral insertion sites, implicating over 300 loci in tumorigenesis. This dataset reveals 20 genes that are specifically mutated in either p19ARF-deficient, p53-deficient or wild-type mice (including Flt3, mmu-mir-106a-363, Smg6, and Ccnd3), as well as networks of significant collaborative and mutually exclusive interactions between cancer genes. Furthermore, we found candidate tumor suppressor genes, as well as distinct clusters of insertions within genes like Flt3 and Notch1 that induce mutants with different spectra of genetic interactions. Cross species comparative analysis with aCGH data of human cancer cell lines revealed known and candidate oncogenes (Mmp13, Slamf6, and Rreb1) and tumor suppressors (Wwox and Arfrp2). This dataset should prove to be a rich resource for the study of genetic interactions that underlie tumorigenesis.

A comparison of genetic network models.

With the completion of the sequencing of the human genome, the need for tools capable of unraveling the interaction and functionality of genes becomes extremely urgent. In answer to this quest, the advent of microarray technology provides the opportunity to perform large scale gene expression analyses. Recently, genetic networks were proposed as a possible methodology for modeling genetic interactions. Since then, a wide variety of different models have been introduced. However, it is, in general, unclear what the strengths and weaknesses of each of these approaches are and where these models overlap and differ. This paper compares different genetic modeling approaches that attempt to extract the gene regulation matrix from expression data. A taxonomy of continuous genetic network models is proposed and the following important characteristics are suggested and employed to compare the models: (1) inferential power; (2) predictive power; (3) robustness; (4) consistency; (5) stability and (6) computational cost. Where possible, synthetic time series data are employed to investigate some of these properties.

De novo sequencing, assembly and analysis of the genome of the laboratory strain Saccharomyces cerevisiae CEN.PK113-7D, a model for modern industrial biotechnology.

Saccharomyces cerevisiae CEN.PK 113-7D is widely used for metabolic engineering and systems biology research in industry and academia. We sequenced, assembled, annotated and analyzed its genome. Single-nucleotide variations (SNV), insertions/deletions (indels) and differences in genome organization compared to the reference strain S. cerevisiae S288C were analyzed. In addition to a few large deletions and duplications, nearly 3000 indels were identified in the CEN.PK113-7D genome relative to S288C. These differences were overrepresented in genes whose functions are related to transcriptional regulation and chromatin remodelling. Some of these variations were caused by unstable tandem repeats, suggesting an innate evolvability of the corresponding genes. Besides a previously characterized mutation in adenylate cyclase, the CEN.PK113-7D genome sequence revealed a significant enrichment of non-synonymous mutations in genes encoding for components of the cAMP signalling pathway. Some phenotypic characteristics of the CEN.PK113-7D strains were explained by the presence of additional specific metabolic genes relative to S288C. In particular, the presence of the BIO1 and BIO6 genes correlated with a biotin prototrophy of CEN.PK113-7D. Furthermore, the copy number, chromosomal location and sequences of the MAL loci were resolved. The assembled sequence reveals that CEN.PK113-7D has a mosaic genome that combines characteristics of laboratory strains and wild-industrial strains.