Research Outputs
Permanent URI for this communityhttps://hdl.handle.net/20.500.14288/2
Browse
7 results
Search Results
Publication Metadata only A multitask multiple kernel learning formulation for discriminating early- and late-stage cancers(Oxford University Press (OUP), 2020) N/A; N/A; Department of Industrial Engineering; Rahimi, Arezou; Gönen, Mehmet; PhD Student; Faculty Member; Department of Industrial Engineering; Graduate School of Sciences and Engineering; College of Engineering; N/A; 237468Motivation: Genomic information is increasingly being used in diagnosis, prognosis and treatment of cancer. The severity of the disease is usually measured by the tumor stage. Therefore, identifying pathways playing an important role in progression of the disease stage is of great interest. Given that there are similarities in the underlying mechanisms of different cancers, in addition to the considerable correlation in the genomic data, there is a need for machine learning methods that can take these aspects of genomic data into account. Furthermore, using machine learning for studying multiple cancer cohorts together with a collection of molecular pathways creates an opportunity for knowledge extraction. Results: We studied the problem of discriminating early- and late-stage tumors of several cancers using genomic information while enforcing interpretability on the solutions. To this end, we developed a multitask multiple kernel learning (MTMKL) method with a co-clustering step based on a cutting-plane algorithm to identify the relationships between the input tasks and kernels. We tested our algorithm on 15 cancer cohorts and observed that, in most cases, MTMKL outperforms other algorithms (including random forests, support vector machine and single-task multiple kernel learning) in terms of predictive power. Using the aggregate results from multiple replications, we also derived similarity matrices between cancer cohorts, which are, in many cases, in agreement with available relationships reported in the relevant literature.Publication Metadata only DeepCOP: deep learning-based approach to predict gene regulating effects of small molecules(Oxford Univ Press, 2020) Woo, Godwin; Fernandez, Michael; Hsing, Michael; Cherkasov, Artem; N/A; N/A; Lack, Nathan Alan; Cavga, Ayşe Derya; Faculty Member; PhD Student; School of Medicine; Graduate School of Sciences and Engineering; 120842; N/AMotivation: Recent advances in the areas of bioinformatics and chemogenomics are poised to accelerate the discovery of small molecule regulators of cell development. Combining large genomics and molecular data sources with powerful deep learning techniques has the potential to revolutionize predictive biology. In this study, we present Deep gene COmpound Profiler (DeepCOP), a deep learning based model that can predict gene regulating effects of low-molecular weight compounds. This model can be used for direct identification of a drug candidate causing a desired gene expression response, without utilizing any information on its interactions with protein target(s). Results: In this study, we successfully combined molecular fingerprint descriptors and gene descriptors (derived from gene ontology terms) to train deep neural networks that predict differential gene regulation endpoints collected in LINCS database. We achieved 10-fold cross-validation RAUC scores of and above 0.80, as well as enrichment factors of >5. We validated our models using an external RNA-Seq dataset generated in-house that described the effect of three potent antiandrogens (with different modes of action) on gene expression in LNCaP prostate cancer cell line. The results of this pilot study demonstrate that deep learning models can effectively synergize molecular and genomic descriptors and can be used to screen for novel drug candidates with the desired effect on gene expression. We anticipate that such models can find a broad use in developing novel cancer therapeutics and can facilitate precision oncology efforts.Publication Metadata only Discriminating early- and late-stage cancers using multiple kernel learning on gene sets(Oxford Univ Press, 2018) N/A; N/A; Department of Industrial Engineering; Rahimi, Arezou; Gönen, Mehmet; PhD Student; Faculty Member; Department of Industrial Engineering; Graduate School of Sciences and Engineering; College of Engineering; N/A; 237468Motivation: Identifying molecular mechanisms that drive cancers from early to late stages is highly important to develop new preventive and therapeutic strategies. Standard machine learning algorithms could be used to discriminate early-and late-stage cancers from each other using their genomic characterizations. Even though these algorithms would get satisfactory predictive performance, their knowledge extraction capability would be quite restricted due to highly correlated nature of genomic data. That is why we need algorithms that can also extract relevant information about these biological mechanisms using our prior knowledge about pathways/gene sets. Results: In this study, we addressed the problem of separating early- and late-stage cancers from each other using their gene expression profiles. We proposed to use a multiple kernel learning (MKL) formulation that makes use of pathways/gene sets (i) to obtain satisfactory/improved predictive performance and (ii) to identify biological mechanisms that might have an effect in cancer progression. We extensively compared our proposed MKL on gene sets algorithm against two standard machine learning algorithms, namely, random forests and support vector machines, on 20 diseases from the Cancer Genome Atlas cohorts for two different sets of experiments. Our method obtained statistically significantly better or comparable predictive performance on most of the datasets using significantly fewer gene expression features. We also showed that our algorithm was able to extract meaningful and disease-specific information that gives clues about the progression mechanism.Publication Metadata only Folding dynamics of proteins from denatured to native state: principal component analysis(Mary Ann Liebert, Inc, 2004) N/A; Department of Chemical and Biological Engineering; Department of Computer Engineering; Department of Chemical and Biological Engineering; Department of Chemical and Biological Engineering; Palazoğlu, Ahmet; Gürsoy, Attila; Arkun, Yaman; Erman, Burak; Other; Faculty Member; Faculty Member; Faculty Member; Department of Computer Engineering; Department of Chemical and Biological Engineering; College of Engineering; College of Engineering; College of Engineering; College of Engineering; N/A; 8745; 108526; N/ASeveral trajectories starting from random configurations and ending in the native state for chymotrypsin inhibitor 2, CI2, are generated using a Go-type model where the backbone torsional angles execute random jumps on which a drift towards their native values is superposed. Bond lengths and bond angles are kept fixed, and the size of the backbone atoms and side groups are recognized. The large datasets obtained are analyzed using a particular type of principal component analysis known as Karhunen - Loeve expansion (KLE). Trajectories are decomposed separately into modes in residue space and time space. General features of different folding trajectories are compared in the modal space and relationships between the structure of CI2 and its folding dynamics are obtained. Dynamic scaling and order reduction of the folding trajectories are discussed. A continuous wavelet transform is used to decompose the nonstationary folding trajectories into windows exhibiting different features of folding dynamics. Analysis of correlations confirms the known two-state nature of folding of CI2. All of the conserved residues of the protein are shown to be stationary in the small modes of the residue space. The sequential nature of folding is shown by examining the slow modes of the trajectories. The present model of protein folding dynamics is compared with the simple Rouse model of polymer dynamics. Principal component analysis is shown to be a very effective tool for the characterization of the general folding features of proteins.Publication Metadata only Path2Surv: Pathway/gene set-based survival analysis using multiple kernel learning(Oxford University Press (OUP), 2019) N/A; Department of Industrial Engineering; Department of Industrial Engineering; Dereli, Onur; Oğuz, Ceyda; Gönen, Mehmet; PhD Student; Faculty Member; Faculty Member; Department of Industrial Engineering; Graduate School of Sciences and Engineering; College of Engineering; College of Engineering; N/A; 6033; 237468Motivation: Survival analysis methods that integrate pathways/gene sets into their learning model could identify molecular mechanisms that determine survival characteristics of patients. Rather than first picking the predictive pathways/gene sets from a given collection and then training a predictive model on the subset of genomic features mapped to these selected pathways/gene sets, we developed a novel machine learning algorithm (Path2Surv) that conjointly performs these two steps using multiple kernel learning. Results: We extensively tested our Path2Surv algorithm on 7655 patients from 20 cancer types using cancer-specific pathway/gene set collections and gene expression profiles of these patients. Path2Surv statistically significantly outperformed survival random forest (RF) on 12 out of 20 datasets and obtained comparable predictive performance against survival support vector machine (SVM) using significantly fewer gene expression features (i.e. less than 10% of what survival RF and survival SVM used).Publication Metadata only Prediction of protein-protein interactions by combining structure and sequence conservation in protein interfaces(Oxford Univ Press, 2005) N/A; Department of Computer Engineering; Department of Chemical and Biological Engineering; Aytuna, Ali Selim; Gürsoy, Attila; Keskin, Özlem; Master Student; Faculty Member; Faculty Member; Department of Computer Engineering; Department of Chemical and Biological Engineering; Graduate School of Sciences and Engineering; College of Engineering; College of Engineering; N/A; 8745; 26605Motivation: Elucidation of the full network of protein- protein interactions is crucial for understanding of the principles of biological systems and processes. Thus, there is a need for in silico methods for predicting interactions. We present a novel algorithm for automated prediction of protein-protein interactions that employs a unique bottom-up approach combining structure and sequence conservation in protein interfaces. Results: Running the algorithm on a template dataset of 67 interfaces and a sequentially non-redundant dataset of 6170 protein structures, 62616 potential interactions are predicted. These interactions are compared with the ones in two publicly available interaction databases (Database of Interacting Proteins and Biomolecular Interaction Network Database) and also the Protein Data Bank. A significant number of predictions are verified in these databases. The unverified ones may correspond to (1) interactions that are not covered in these databases but known in literature, (2) unknown interactions that actually occur in nature and (3) interactions that do not occur naturally but may possibly be realized synthetically in laboratory conditions. Some unverified interactions, supported significantly with studies found in the literature, are discussed.Publication Metadata only Segregation indices for disease clustering(Wiley-Blackwell, 2014) Department of Mathematics; Ceyhan, Elvan; Faculty Member; Department of Mathematics; College of Sciences; N/ASpatial clustering has important implications in various fields. In particular, disease clustering is of major public concern in epidemiology. In this article, we propose the use of two distance-based segregation indices to test the significance of disease clustering among subjects whose locations are from a homogeneous or an inhomogeneous population. We derive the asymptotic distributions of the segregation indices and compare them with other distance-based disease clustering tests in terms of empirical size and power by extensive Monte Carlo simulations. The null pattern we consider is the random labeling (RL) of cases and controls to the given locations. Along this line, we investigate the sensitivity of the size of these tests to the underlying background pattern (e.g., clustered or homogenous) on which the RL is applied, the level of clustering and number of clusters, or to differences in relative abundances of the classes. We demonstrate that differences in relative abundances have the highest influence on the empirical sizes of the tests. We also propose various non-RL patterns as alternatives to the RL pattern and assess the empirical power performances of the tests under these alternatives. We observe that the empirical size of one of the indices is more robust to the differences in relative abundances, and this index performs comparable with the best performers in literature in terms of power. We illustrate the methods on two real-life examples from epidemiology. Copyright (c) 2013 John Wiley & Sons, Ltd.