Radiotherapy remains one of the most important treatment modalities for cancer therapy. Malignant tumors show an extended spectrum of intrinsic radiation sensitivity even among tumors of the same entity or with similar histological features. Predicting intrinsic radiation sensitivity might improve treatment outcome and allow individualized treatment. Hence, an assay that provides a predictive information of the tumor’s intrinsic radiation sensitivity is of great need. Histone H2AX, a histone variant of histone H2A family, is rapidly phosphorylated upon DNA double strand break (DSB) induction resulting in gamma H2AX (γH2AX). Gamma H2AX accumulates at DNA DSB sites and subsequently recruits DNA damage repair factors. Formation of γH2AX is visualized by an immunohistology-based approach and detected as foci under an epifluorescent microscope. Gamma H2AX foci represent DNA DSBs, while residual γH2AX foci (foci detected 24 h post irradiation) are considered as unrepaired damages. In previous studies, the γH2AX foci assay showed a high potential as a predictive method for radiosensitivity. This thesis aims to further translate and optimize the ex vivo γH2AX foci assay for a clinical applicability. In this study, all experiments were performed using human head and neck squamous cell carcinoma (hHNSCC) xenograft models. For ex vivo investigations, tumors on the hind legs of nude mice were excised and cut into multiple pieces, or fine-needle biopsies of the tumors were taken. Tumor biopsies were reoxygenated in culture medium for 10 h or 24 h followed by radiation exposure of 0 8 Gy. Tumor biopsies were fixed and embedded in paraffin 24 h post irradiation. For the γH2AX foci assay under in vivo conditions, tumors-bearing mice were irradiated with single doses of 0 8 Gy. Tumors were excised, fixed, and paraffin embedded 24 h post irradiation. Manual quantification of γH2AX foci was performed exclusively in perfused areas, which were identified by pimonidazole (hypoxic marker) and BrdU (proliferation marker) staining. Foci number was corrected, normalized, and statistically analyzed by a linear mixed effects model (LMEM), linear regression model or analysis of covariance. To investigate tumor heterogeneity in the ex vivo γH2AX foci assay, γH2AX foci were enumerated in four equally treated tumor specimens per group i.e. unirradiated and ex vivo irradiated with 4 Gy. Strong intratumoral heterogeneity in γH2AX foci was determined with a minor intertumoral heterogeneity. No significant effect of reoxygenation between 10 h or 24 h was observed, enhancing clinical practicability of the assay. The effect of experimental settings was studied by analyzing data from this study (ex vivo) and from comparable published data (in vivo) with LMEM. Radiation induced nuclear area alteration was detected in some of the evaluated tumor models in under both experimental conditions. A greater intra and intertumoral heterogeneity were observed in the ex vivo set up compared to the in vivo set up. Radiation response determined by the γH2AX foci assay in ex vivo irradiated biopsies and in the corresponding in vivo irradiated tumors was evaluated. Between in vivo and ex vivo, four out of five tumor models showed comparable slopes of dose response curves (SDRC) of normalized and corrected γH2AX foci. SDRC of normalized γH2AX foci was able to classify tumors according to their intrinsic radiation sensitivity (TCD50). In conclusion, the ex vivo γH2AX foci assay holds a promising potential for predicting radiation sensitivity in solid tumors. The comparable radiation response assessed by γH2AX foci of in vivo irradiated tumors and the matching ex vivo irradiated tumor biopsies supports clinical applicability of the assay. Using SDRC of γH2AX foci as a predictor of radiosensitivity, radioresistant and radiosensitive tumors could be classified. The significant intratumoral heterogeneity in the ex vivo γH2AX foci assay suggests a limited representativeness of a single biopsy for radiosensitivity prediction. Additionally, the change of tumor microenvironment modulated cellular adaptation and DNA damage repair capability. The outcomes suggested that a sufficient number of cells, regions of interest, and biopsies are required to obtain a solid prediction.:Contents List of Abbreviations List of Figures List of Tables 1. Introduction 1.1 Effect of ionizing radiation on cellular level 1.1.1 Radiation induces cell death 1.1.2 Cell-cycle arrest mediated by radiation 1.2 DNA damage repair 1.2.1 Non homologous end joining (NHEJ) 1.2.2 Homologous recombination (HR) 1.2.3 Base damage repair and single strand break repair 1.2.4 Role of γH2AX in DNA damage repair 1.3 Prediction of tumor radioresponsiveness 1.3.1 Prediction of tumor radiation sensitivity by γH2AX 2 Tumor heterogeneity determined with a γH2AX foci assay: A study in human head and neck squamous cell carcinoma (hHNSCC) 2.1 Summary of the publication 3 Heterogeneity of γH2AX foci increases in ex vivo biopsies relative to in vivo tumors. 3.1 Summary of the publication 4 Comparable radiation response of ex vivo and in vivo irradiated tumor samples determined by residual γH2AX foci 4.1 Summary of the manuscript 5 Discussion 5.1 Tumor heterogeneity in γH2AX foci assay 5.2 Alteration of nuclear area post irradiation 5.3 Clinical relevance of the γH2AX foci assay 5.4 Technical challenges and limitations of the assay 5.5 Conclusion and Outlook 6 Abstract 7 Zusammenfassung 8 Bibliography Acknowledgement Appendices Part A: Materials A.1 Tumor lines A.2 Chemicals and Materials A.3 Devices and Software Part B: Supplementary materials B.1 Supplementary materials of publication I B.2 Supplementary materials of publication II B.3 Supplementary materials of manuscript

Proteomics as a modern comprehensive approach to the analysis of proteomes was applied in three projects aimed at diagnosis and therapy of cancer. The aim of the first the project was to find a new diagnostic biomarker for ovarian cancer. Two different comparative proteomic approaches were used for comparative analysis of sera from patients diagnosed with ovarian cancer and from healthy age-matched women. We identified -1-antitrypsin with increased concentration in patien sera, and apolipoprotein A4 and retinol-binding protein 4 (RBP4) with significantly decreased concentration in patients. The significantly decerased concentration of RBP4 in patients is a new observation. We propose that RBP4 is either decreased in ovarian cancer patients as a result of its reduced production by ovary or it may reflect less specific systemic changes, for instance early onset of cancer cachexia. The second project was focused on gaining insight into the molecular mechanism of cytarabine resistance in mantle cell lymphoma (MCL). Proteomic and transcriptomic analyses of cytarabine-resistant cells revealed marked downregulation of deoxycytidine kinase (DCK) - a protein essential to intracellular activation of purine and pyrimidine nucleosides and their analogues including cytarabine. The cytarabine-resistant MCL...

In this thesis, we address the identification of biomarkers in high-dimensional omics data. The identification of valid biomarkers is especially relevant for personalized medicine that depends on accurate prediction rules. Moreover, biomarkers elucidate the provenance of disease, or molecular changes related to disease. From a statistical point of view the identification of biomarkers is best cast as variable selection. In particular, we refer to variables as the molecular attributes under investigation, e.g. genes, genetic variation, or metabolites; and we refer to observations as the specific samples whose attributes we investigate, e.g. patients and controls. Variable selection in high-dimensional omics data is a complicated challenge due to the characteristic structure of omics data. For one, omics data is high-dimensional, comprising cellular information in unprecedented details. Moreover, there is an intricate correlation structure among the variables due to e.g internal cellular regulation, or external, latent factors. Variable selection for uncorrelated data is well established. In contrast, there is no consensus on how to approach variable selection under correlation. Here, we introduce a multivariate framework for variable selection that explicitly accounts for the correlation among markers. In particular, we present two novel quantities for variable importance: the correlation-adjusted t (CAT) score for classification, and the correlation-adjusted (marginal) correlation (CAR) score for regression. The CAT score is defined as the Mahalanobis-decorrelated t-score vector, and the CAR score as the Mahalanobis-decorrelated correlation between the predictor variables and the outcome. We derive the CAT and CAR score from a predictive point of view in linear discriminant analysis and regression; both quantities assess the weight of a decorrelated and standardized variable on the prediction rule. Furthermore, we discuss properties of both scores and relations to established quantities. Above all, the CAT score decomposes Hotelling’s T 2 and the CAR score the proportion of variance explained. Notably, the decomposition of total variance into explained and unexplained variance in the linear model can be rewritten in terms of CAR scores. To render our approach applicable on high-dimensional omics data we devise an efficient algorithm for shrinkage estimates of the CAT and CAR score. Subsequently, we conduct extensive simulation studies to investigate the performance of our novel approaches in ranking and prediction under correlation. Here, CAT and CAR scores consistently improve over marginal approaches in terms of more true positives selected and a lower model error. Finally, we illustrate the application of CAT and CAR score on real omics data. In particular, we analyze genomics, transcriptomics, and metabolomics data. We ascertain that CAT and CAR score are competitive or outperform state of the art techniques in terms of true positives detected and prediction error.

Background: N-terminal pro-brain natriuretic peptide (NT-proBNP) is secreted by cardiac ventricular myocytes upon pressure and volume overload and is a prognostic marker to monitor the severity of precapillary pulmonary hypertension and the extent of right heart failure. Objectives: The impact of physical exercise on NT-proBNP levels in patients with left heart disease was demonstrated previously. No data regarding patients with isolated right heart failure and the influence of acute exercise on NT-proBNP serum levels exist. Methods: Twenty patients with precapillary pulmonary hypertension were examined. Hemodynamic parameters were measured during right heart catheterization. Serum NT-proBNP of patients was measured at rest, after a 6-min walking test, during ergospirometry and during recovery, all within 7 h. Significant differences in sequential NT-proBNP values, relative changes compared to values at rest and the correlation between NT-proBNP and obtained parameters were assessed. Results: At rest, the mean serum level of NT-proBNP was 1,278 ± 998 pg/ml. The mean level of NT-proBNP at maximal exercise was increased (1,592 ± 1,219 pg/ml), whereas serum levels decreased slightly during recovery (1,518 ± 1,170 pg/ml). The relative increase of serum NT-proBNP during exercise correlated with pulmonary vascular resistance (r = 0.45; p = 0.026) and cardiac output (r = –0.5; p = 0.015). Conclusions: In this study, we demonstrated acute changes in NT-proBNP levels due to physical exercise in a small group of patients with precapillary pulmonary hypertension. Our results also confirm the predominant usefulness of NT-proBNP as an intraindividual parameter of right heart load. / Dieser Beitrag ist mit Zustimmung des Rechteinhabers aufgrund einer (DFG-geförderten) Allianz- bzw. Nationallizenz frei zugänglich.

Gene expression can serve as a powerful predictor for disease progression and other phenotypes. Consequently, microarrays, which capture gene expression genome-wide, have been used widely over the past two decades to derive biomarker signatures for tasks such as cancer grading, prognosticating the formation of metastases, survival, and others. Each of these signatures was selected and optimized for a very specific phenotype, tissue type, and experimental set-up. While all of these differences may naturally contribute to very heterogeneous and different biomarker signatures, all cancers share characteristics regardless of particular cell types or tissue as summarized in the hallmarks of cancer. These commonalities could give rise to biomarker signatures, which perform well across different phenotypes, cell and tissue types. Here, we explore this possibility by employing a network-based approach for pan-cancer biomarker discovery. We implement a random surfer model, which integrates interaction, expression, and phenotypic information to rank genes by their suitability for outcome prediction. To evaluate our approach, we assembled 105 high-quality microarray datasets sampled from around 13,000 patients and covering 13 cancer types. We applied our approach (NetRank) to each dataset and aggregated individual signatures into one compact signature of 50 genes. This signature stands out for two reasons. First, in contrast to other signatures of the 105 datasets, it is performant across nearly all cancer types and phenotypes. Second, It is interpretable, as the majority of genes are linked to the hallmarks of cancer in general and proliferation specifically. Many of the identified genes are cancer drivers with a known mutation burden linked to cancer. Overall, our work demonstrates the power of network-based approaches to compose robust, compact, and universal biomarker signatures for cancer outcome prediction.

Background: Neuropilin (NRP) is a transmembrane protein, which has been shown to be a pro-angiogenic mediator and implicated as a potential driver of cancer progression. NRP-1 up-regulation in ovarian cancer tissue predicts poor prognosis. However, the clinical relevance of the soluble form of NRP-1 (sNRP-1) as a circulating biomarker in ovarian cancer patients is unknown. - Methods/patients cohort: sNRP-1 levels were quantified in a cohort of 88 clinically documented ovarian cancer patients by a commercially available sNRP-1 enzyme-linked immunosorbent assay (ELISA) kit (Biomedica, Vienna, Austria). Patients (81.8% with FIGOIII/IV) received primary cytoreductive surgery with the aim of macroscopic complete resection (achieved in 55.7% of patients) and the recommendation of adjuvant chemotherapy in line with national guidelines. - Results: Higher levels of sNRP-1 reflected more advanced disease (FIGO III/IV) and indicated a trend towards suboptimal surgical outcome, i.e. any residual tumor. sNRP-1 was neither related to the patients’ age nor the BRCA1/2 mutational status. Patients with higher sNRP-1 levels at primary diagnosis had a significantly reduced progression-free survival (PFS) (HR = 0.541, 95%CI: 0.304 - 0.963; p = 0.037) and overall survival (OS) (HR = 0.459, 95%CI: 0.225 - 0.936; p = 0.032). Principal component analysis showed that sNRP-1 levels were unrelated to the circulating hepatocyte growth factor (HGF) and the soluble ectodomain of its receptor the tyrosine kinase mesenchymal–epithelial transition (c-MET), suggesting that there is no proportional serological concentration gradient of soluble components of the NRP-1/HGF/c-MET signaling axis. - Conclusions: In line with the previously shown tissue-based prognostic role, we demonstrated for the first time that sNRP-1 can also act as a readily accessible, prognostic biomarker in the circulation of patients with ovarian cancer at primary diagnosis. Given its known role in angiogenesis and conferring resistance to the poly ADP-ribose polymerase (PARP) inhibitor olaparib in vitro, our results encourage more detailed investigation into sNRP-1 as a potential predictive biomarker for bevacizumab and/or PARP-inhibitor treatment.

Im Vergleich zu anderen Krebsentitäten, konnten Patienten mit PDAC bisher kaum von Therapieerfolgen der Präzisionsmedizin profitieren. Um diese Problematik zu adressieren, habe ich eine umfassende molekularbiologische Studie durchgeführt, um prädiktive Biomarker zu identifizieren und die Risikostratifizierung der Patienten zu verfeinern. Mittels gen-spezifischer Sequenzierung und gezielter RNA-Expressionsanalyse wurden 293 R0-resezierte Patienten aus einer multizentrischen Phase-III-Studie untersucht. Ziel der klinischen Studie war der Vergleich von adjuvanter Chemotherapie mit Gemcitabin entweder mit oder ohne Zusatz von Erlotinib. Für meine Arbeit wurden die Patientenproben unter Verwendung einer nicht-negativen Matrixfaktorisierung (NMF) basierend auf ihren Einzelnukleotidvarianten (SNV) und ihren Kopienzahlveränderungen (CNA) gruppiert und auf klinische und molekularbiologische Unterschiede untersucht. Um die biologischen Hintergründe der identifizierten genetischen Besonderheiten zu verstehen, wurden anschließend Zelllinien genetisch modifiziert und in vitro modelliert. Es wurden 1086 SNVs und 4157 CNAs identifiziert. Dabei wiesen 99% aller Patienten mindestens eine genetische Veränderung auf, mit durchschnittlich 18 Aberrationen pro Patient. In Übereinstimmung mit früheren Berichten waren KRAS, TP53, CDKN2A und SMAD4 die am häufigsten betroffenen Gene. Alterationen in diesen Genen konnten in 63-93 % der Fälle nachgewiesen werden. Basierend darauf konnte ich fünf Patientengruppen identifizieren die sich in ihren biologischen Charakteristika unterscheiden und Angriffspunkte für gezielte Therapien bieten. Mittels NMF wurden zudem SMAD4alt MAPK9low als prognostische Biomarker für Erlotinib identifiziert. Anschließende in vitro Experimente zeigten, dass dies nicht auf eine Erhöhung der Erlotinib-Zelltoxizität zurückzuführen ist. Zuletzt definiere ich einen prognostischen Score der genutzt werden kann um das Überleben von R0-resizierten PDAC Patienten abzuschätzen. / In contrast to other cancer entities, PDAC patients have not benefited from recent improvements in precision medicine. To address this gap, I embarked on a comprehensive molecular study to identify predictive biomarkers and refine risk stratification. I performed targeted sequencing and targeted RNA expression analysis of 293 R0-resected patients from a multicenter phase III trial comparing adjuvant chemotherapy of gemcitabine with or without erlotinib. Patients were clustered using non-negative matrix factorization (NMF) based on their single nucleotide variant (SNV) and copy number alteration (CNA) statuses. Overall (OS) and disease-free survival (DFS) were analysed with the multivariate cox hazard and log rank tests. Finally, using a method based on CRISPR/Cas, findings from the patient cohort where modeled in vitro to assess their biological backgrounds. A total of 1,086 SNVs and 4,157 CNAs were found with at least one genetic alteration in 99% of all patients, and an average of 18 aberrations per patient. In line with previous reports, KRAS, TP53, CDKN2A, and SMAD4 were the most frequently affected genes, detected in 63–93 % of cases. In this thesis, I identified five biologically distinct patient subgroups with different actionable lesions that may serve for refined PDAC classification and tailored treatment approaches. NMF based clustering and subsequent differential expression analysis revealed SMAD4alt (SNV and/or CAN in SMAD4) MAPK9low (MAPK9 expression below median) as prognostic erlotinib biomarker. Modeling of SMAD4alt MAPK9low status in vitro showed that the effect is not based on increased erlotinib toxicity. Finally, I proposed a genetic risk score for prognostic evaluation of newly diagnosed R0-resected PDAC patients.

Das metastasierende kolorektale Karzinom stellt eine große Herausforderung in der Krebstherapie dar. Verlässliche und effiziente Biomarker zur Prognose des Krankheitsverlaufes oder der Therapieantwort (Prädiktion) sind rar. Metastasis-associated in colon cancer 1 (MACC1) ist ein prognostischer, prädiktiver und kausaler Biomarker für verschiedene Tumorentitäten. Durch die Induzierung von Zielgenen, wie z.B. MET, beeinflusst es Signalwege wie MEK/ERK und AKT/β-catenin und fördert so Zellproliferation und -motilität sowie Tumorprogression und Metastasierung in vivo. Diese Arbeit sollte neue Strategien erforschen diese Prozesse durch die Inhibition von MACC1 auf Transkriptions- und Signaltransduktionsebene zu unterbinden. Mit zwei verschiedenen Screeningmethoden konnten wir Statine als potente transkriptionelle Inhibitoren von MACC1 als auch phosphotyrosin (pY)-abhängige Interaktionen von MACC1 mit essentiellen Signalmolekülen identifizieren: SHP2, GRB2, SHC1, PLCG1 und STAT5B. Statine verringerten MACC1-spezifische Proliferation und Koloniebildung in vitro als auch Tumor Wachstum und Metastasierung in vivo bei Dosen äquivalent der humanen Standardtherapie zur Blutlipidsenkung. Mutation der pY-Bindungsstellen reduzierte die Aktivität des MACC1-induzierten ERK Signalwegs sowie Zellmigration und -proliferation. Anhand unserer Daten orchestriert MACC1, abhängig von MET und EGFR, neue SHP2/SRC/ERK und PKA/SRC/CREB Signalkaskaden zu einem malignen Phänotyp. Gezielte Intervention restringierte die MACC1-abhängige Koloniebildung, was neue therapeutische Interventionspunkte identifiziert und eine hervorragende Basis für Untersuchungen zur Kombinationstherapie darstellt. Die weitere Erforschung der spatiotemporalen Organisation des MACC1 Signalosoms und assoziierter Signalkaskaden soll das volle therapeutische Potential von MACC1 ausschöpfen. Wir empfehlen zudem Statine in der Krebstherapie bzw. -prävention, besonders bei MACC1-stratifzierten Patienten, anzuwenden. / Metastatic colorectal cancer still represents a major challenge in therapy. Reliable and efficient biomarkers for early prognosis of disease course or treatment response (prediction) remain scarce. Metastasis-associated in colon cancer 1 (MACC1) has been established as prognostic, predictive and causal biomarker for several tumor entities. Its induction of target genes such as MET affects several signaling pathways including MEK/ERK and AKT/β-catenin. Thus, it promotes cellular proliferation and motility as well as tumor progression and metastasis formation in vivo. This study intended to explore new strategies to inhibit these processes by targeting MACC1 on transcriptional and signaling level. By two distinct screening methods, we identified statins as potent MACC1 transcriptional inhibitors as well as phosphotyrosine (pY)-dependent interactions of MACC1 with crucial signaling molecules: SHP2, GRB2, SHC1, PLCG1 and STAT5B. Statins showed MACC1-specific reduction of proliferation and colony formation in vitro as well as restriction of tumor growth and metastasis formation in vivo at doses equivalent to human standard lipid reduction therapy. Mutation of the pY-interaction sites abrogated MACC1-dependent ERK signaling as well as cell migration and proliferation. Our data further suggest that MACC1 governs SHP2/SRC/ERK and PKA/SRC/CREB axes conferring a malignant phenotype in response to MET and EGFR. Targeted intervention restricted MACC1-dependent colony formation which indicates new drug intervention points for MACC1 signaling and provides an excellent baseline for further investigations of combinatorial treatments. Additional research about the spatiotemporal organization of MACC1 signalosome formation and downstream signaling will reveal the entire potential of MACC1 as therapeutic target, whereas statins should already be considered for cancer therapy or prevention, especially in patients stratified for MACC1 expression.

Das Entwicklungsstadium bestimmter Krankheiten ist eng mit der Konzentration diverser Proteine in biologischen Proben verknüpft. Eine sensitive Detektion dieser sogenannten Biomarker kann somit maßgeblich zu einer frühzeitigen Diagnose beitragen. In der vorliegenden Arbeit wurden strukturierte, fluorogene Sonden entwickelt, die die Möglichkeit bieten in einem homogenen Verfahren Zielproteine sensitiv, direkt und in Echtzeit nachzuweisen. Die peptidische Erkennungssequenz für das Zielprotein wurde dabei von zwei zueinander komplementären PNA-Segmenten flankiert. Die Sonden besaßen dadurch eine haarnadelförmige Anordnung, die kontrolliert eingebaute Reporter in eine enge Proximität zwang und ein minimales Hintergrundsignal im ungebundenen Zustand gewährleistete. Durch die Wechselwirkung mit dem Zielprotein erfolgte eine Reorganisation der Sondenstruktur, die fluoreszenzspektroskopisch verfolgt werden konnte. Für den Einbau der fluorogenen Einheiten wurden verschiedene Strategien entwickelt und die resultierenden Architekturen bzgl. ihres Einsatzes als sensitives Detektionssystem validiert. Als Zielproteine wurden die intrazellulären SH2-Domänen der Src- und Lck-Kinase sowie die extrazelluläre Matrix-Metalloprotease MMP-7, ein proteolytischer Biomarker für Krebs, untersucht. Besonders die neuartigen In-Stem Hairpin Peptide Beacons (IS-HPBs), bei denen fluorogene Pseudonukleobasen in die PNA-Stammregion eingebaut wurden, zeichneten sich als sensitive Proteasereporter mit einer bis zu 50-fachen Signalverstärkung aus. Mit einem excimerbasierten IS-HPB und einer zeitaufgelösten Fluoreszenzmethode konnte die direkte Detektion von MMP-7 bei einer kritischen Konzentration von 1 nM im humanen Blutserum erreicht werden. Eine mögliche Anwendbarkeit in der medizinischen Diagnostik wurde somit bekräftigt. Weiterhin wurden erste Hinweise mithilfe thermodynamischer Untersuchungen erhalten, dass die Strukturierung einer peptidischen Sonde zu einer erhöhten Selektivität beiträgt. / The developmental stage of certain diseases is closely linked to the concentration of various proteins in biological samples. A sensitive detection of these so-called biomarkers can thus significantly contribute to an early diagnosis. In the present work, structured, fluorogenic probes were developed that offer the possibility of a sensitive, direct and in real-time detection of target proteins in a homogeneous process. The peptidic recognition sequence for the target protein was thereby flanked by two self-complementary PNA segments. As a result, the probes possessed a hairpin-type arrangement, in which suitable appended labels are forced into close proximity and guaranteed a minimal background signal in the unbound state. By interacting with the target protein a reorganization of the probe structure occured, which could be followed by fluorescence spectroscopy. To embed the fluorogenic units different approaches were developed and the resulting architectures were validated relating to their use as a sensitive detection system. As target proteins the intracellular SH2-domains of the Src and Lck kinase and the extracellular matrix metalloprotease MMP-7, a proteolytic biomarker for cancer, were investigated. In particular, the new In-Stem Hairpin Peptide Beacons (IS-HPBs), in which fluorogenic pseudo nucleic acids were incorporated into the PNA-stem region, proved as sensitive protease reporters with an up to 50-fold signal amplification. By using an excimer-signaling IS-HPB and a time-resolved fluorescence method the direct detection of MMP-7 with a critical concentration of 1 nM within complex human blood serum was achieved. A possible application in medical diagnostics was thus confirmed. Furthermore, initial indications were obtained using thermodynamic studies that the structure of a peptide-based probe contributes to increased selectivity.

<p>The accumulation of high-throughput data from vast sources has drawn a lot attentions to develop methods for extracting meaningful information out of the massive data. More interesting questions arise from how to combine the disparate information, which goes beyond modeling sparsity and dimension reduction. This dissertation focuses on the innovations in the area of heterogeneous data integration.</p><p>Chapter 1 contextualizes this dissertation by introducing different aspects of meta-analysis and model frameworks for high-dimensional genomic data.</p><p>Chapter 2 introduces a novel technique, joint Bayesian sparse factor analysis model, to vertically integrate multi-dimensional genomic data from different platforms. </p><p>Chapter 3 extends the above model to a nonparametric Bayes formula. It directly infers number of factors from a model-based approach.</p><p>On the other hand, chapter 4 deals with horizontal integration of diverse gene expression data; the model infers pathway activities across various experimental conditions. </p><p>All the methods mentioned above are demonstrated in both simulation studies and real data applications in chapters 2-4.</p><p>Finally, chapter 5 summarizes the dissertation and discusses future directions.</p> / Dissertation