· · · Institutional Repository

Expansion of an existing visualization tool for genomic data.

In this investigation a gene set–based approach is used to compare T-ALL samples to a series of chronologically ordered T-cell development stages. These results revealed two distinct subtypes of T-ALL that can be defined by different cells of origin; either a multipotent hematopoietic stem cell, or a committed T-cell. Subsequently, two-step progression pathways in terms of biological process perturbations were determined for each of these subtypes based on differentially expressed gene sets. The two-step model allowed the characterization of a novel cell type, the T-ALL stem cell. The results indicate that T-ALL can emerge from both multipotent cells as well as committed T-cells, and that the progression pathways of each of these subtypes proceeds via different T-ALL stem cells.

In this thesis an approximation method is discussed that provides similar results to leave-one-out cross-validation but is less time-consuming. By means of this approximation method, estimating the optimal values of ridge and lasso parameters will take less time and carrying out (an approximated version of) double LOOCV will become practically feasible. The method can be used in generalized linear models as well as in Cox' proportional hazards model. In order to show its usefulness, the method is tested on several microarray data sets.

In this thesis we present the database security mechanism and Web services implementation for the Ibidas system. The security mechanism restricts access to data by source per user. Web services provide an interface to call Ibidas methods and share data from and with other applications. We will see the need for the features presented, how they are implemented and the choices which were made during implementation. Furthermore we will show an example of the use of the Web services using the Taverna workbench application to solve a typical situation a researcher could face. We conclude this thesis with a discussion regarding our implementation and some future work indications.

Unbalanced transaction can lead to addition and deletions in genes, which can be an indication of tumor cells. This is measured with array Comparative Genomic Hybridization. To find co-occurring aberrations in DNA, an algorithm was designed. However the execution takes days to find these DNA aberrations. This thesis proposes a partial FPGA based design were the number of parallel computations can be increased. The FPGA communicates with a computer on a gigabit Ethernet, where on the FPGA a hardware based Ethernet controller is build. This design is scalable for FPGA’s, so its performance is linear to the size of the FPGA resources. On a XC4VFX12 device, a minimum speedup of a factor 3 and a maximum speedup of several hundreds is achieved.

In this thesis we present the database security mechanism and Web services implementation for the Ibidas system. The security mechanism restricts access to data by source per user. Web services provide an interface to call Ibidas methods and share data from and with other applications. We will see the need for the features presented, how they are implemented and the choices which were made during implementation. Furthermore we will show an example of the use of the Web services using the Taverna workbench application to solve a typical situation a researcher could face. We conclude this thesis with a discussion regarding our implementation and some future work indications.

Motivation: Synthetic biology aims at building biological systems for useful purposes. Relatively simple gene networks have been engineered, but the design process is limited. Many papers advocate the use of engineering concepts like standardization and modular design to simplify the design process and enable the design of more complex systems. Currently, there are no tools available that implement both concepts in a practical way. Results: We have developed a software tool to show how standardization and modular design can be used for the design of logic gene networks. We introduce gene network templates to be able to use modular design in a practical way and use a standard model to simplify the design process and enable reuse of parameters. We have designed three logic gate templates and used them to build two logic gene networks: a demultiplexer and a D-latch. Our software tool was used to turn the templates into devices and to evaluate the performance of the devices. The results show that the devices are evaluated correctly. Furthermore, the results show that for the design of a gene network our method can be used to indicate which biological parts are preferred at what location in the network.

Automated landmark detection may prove important for the examination and automatic analysis of real-time three-dimensional (3D) echocardiograms. By detecting 3D anatomical landmark points, the standard anatomical views can be extracted automatically in 3D ultrasound images of left ventricle, for better standardization and objective diagnosis. Furthermore, the landmarks can serve as an initialization for other analysis methods, such as segmentation. In this thesis we describe an algorithm that iteratively applies landmark detection in perpendicular planes of the 3D dataset. The landmark detection exploits a large database of expert annotated images, using an extensive set of Haar wavelet-like features for classification, resulting in fast detection times suitable for real-time applications. The detection is performed using two cascades of Adaboost classifiers, that work in different 2D planes, in a coarse to fine scheme. The method is evaluated by measuring the total detection error for the landmark points between the detected positions and the manual ones.

Continuous variable selection using shrinkage procedures have recently been considered as favorable models in a wide range of scientific research; in particular biomedical research. In some cases, it is desirable to select as few predictors as possible, to increase the interpretability of the attained prediction rule. One frequently used shrinkage procedure; the Lasso, imposes a L1 regularization on the regression coefficients of general linear models, inherently leading to sparse prediction rules. When dealing with multi-class prediction in generalized linear models each predictor has a regression coefficient for each class. A major disadvantage is that the Lasso selects individual regression coefficients instead of the more logical selection of predictors. In this paper, we demonstrate a new regularization procedure, based on the Group Lasso in multinomial logistic regression. This results in a lower number of retained predictors, but with similar prediction accuracy when compared to the regular Lasso regularization. To illustrate the new regularization applicability we have employed it on a large cohort of acute myeloid leukemia patients (AML, n=531) who are characterized on a gene expression microarray.

This report is made as part of the Master's Thesis project of the master Computer Science, track Bioinformatics at the Delft University of technology. The main focus of this document lies on the paper that is written as the result of my research on the detection of copy number variations using next generation sequencing data. In the future it is likely that next generation sequencing techniques will replace the current array-comparative genomic hybridization technique that is currently used in clinics. For next generation sequencing data to replace this technique, the minimal coverage required for competitive detection should be known; this was the focus of my work. Besides the main paper, a supplement is provided to give some additional information about the research that was done. A work document is also included, in this document the progress and observations made during the project are registered. The Master's thesis project was done at the Bioinformatics Lab at Delft University of Technology in collaboration with the department of Clinical Genetics of the VU Medical Center in Amsterdam.

In this project we tried to answer the question whether it is possible to obtain in vivo quantitative spatiotemporal gene expression data of mice, by making use of the GFP protein and Biofluorence Imaging. The research was driven by a more specific question, being whether it is possible to detect if metastasis to the bone has occurred in cancer progression studies, using GFP cancel cell lines, i.e cancer cells that produce Green Fluorescent Proteins (GFP), at a continuous known rate. We tried to give generic answers by researching the following two subquestions. 1.) Is it possible to register a 3D mouse atlas to 2D Bioluminescense or Fluorescence photographs, based on only those photographs? 2.) Can we make qualitative statements on the location of gene expression, after registration with an atlas? The first question is covered in a paper called: 'Atlas Driven Registration of 3D Voxel Data to Multi-view Photographs Based on 3D Distance Maps.' It explains a method to register a 3D piecewise deformable mouse atlas to 2 or more photographic sideviews of a mouse. Based on a distance map that we generated from multiple backprojections of the sideview, we were able to construct an energy function that resembled a 'goodness of fit' of the registration. Comparisons were made with a gold standard and we obtained good results with our method. The second question is covered in a paper called: 'Testing for Spatial Gene Enrichment in C. Elegans Using Chronograms and a 1D Worm Atlas'. In this paper we did not focus on the registration process, because this was a straightforward procedure. With an atlas registered to an expression dataset, we applied different statistical tests to answer the question that given the atlas and the expression profile, the observed expression shows enrichment in a selected organ or not. In this paper we show that we are able to filter highly enriched signals out of our complete dataset. We further discuss the added value of our atlas since it is difficult to validate the obtained results of our tests.

Motivation: Numerous genetic disorders can be detected in prenatal diagnosis using Chorionic Villus Sampling. As this brings a 1:100 chance for miscarriage, this method is only applied when the fetus is suspected to have an increased chance at having an aberration. It is known that a small percentage of DNA in maternal plasma is of fetal origin and with the upswing of Next Generation Sequencing (NGS) it is now possible to diagnose trisomy 21 cases non-invasively using maternal blood samples. Other and smaller aberrations can be detected with this kind of blood samples using high coverage NGS, but this makes these methods more expensive. Our goal was to find a method to detect small aberrations using only shallow sequencing (low coverage NGS), keeping the costs per sample to a minimum. Methods: We developed an approach (called WISECONDOR) that determines whether a small region is aberrated by looking whether the read count within that region differs abnormally compared to the read counts of other similarly behaving regions within the same test sample. The other similarly behaving regions are considered to have similar NGS biases, such as GC-richness, and are found by comparing read count correlations over a set of healthy samples. Results: The developed script was able to correctly identify all tested T13, T18 and T21 cases. Aberrations such as del(13)(q12.3q14.3), i(12p10) and i(18q10) were also correctly called. The amount of false positives was minimal and even zero when the size of the aberrations was at least 20Mb, while the test samples varied from only 0.15 to 1.66 fold coverage. This shows that most fetal fetal copy number aberrations can be detected by shallow sequencing of maternal blood samples.