Witold R. Rudnicki

Evaluation of methods for modeling transcription factor sequence specificity

Genomic analyses often involve scanning for potential transcription factor (TF) binding sites using models of the sequence specificity of DNA binding proteins. Many approaches have been developed to model and learn a proteins DNA-binding specificity, but these methods have not been systematically compared. Here we applied 26 such approaches to in vitro protein binding microarray data for 66 mouse TFs belonging to various families. For nine TFs, we also scored the resulting motif models on in vivo data, and found that the best in vitro–derived motifs performed similarly to motifs derived from the in vivo data. Our results indicate that simple models based on mononucleotide position weight matrices trained by the best methods perform similarly to more complex models for most TFs examined, but fall short in specific cases (<10% of the TFs examined here). In addition, the best-performing motifs typically have relatively low information content, consistent with widespread degeneracy in eukaryotic TF sequence preferences.

An efficient implementation of Smith Waterman algorithm on GPU using CUDA, for massively parallel scanning of sequence databases

The Smith Waterman algorithm for sequence alignment is one of the main tools of bioinformatics. It is used for sequence similarity searches and alignment of similar sequences. The high end Graphical Processing Unit (GPU), used for processing graphics on desktop computers, deliver computational capabilities exceeding those of CPUs by an order of magnitude. Recently these capabilities became accessible for general purpose computations thanks to CUDA programming environment on Nvidia GPUs and ATI Stream Computing environment on ATI GPUs. Here we present an efficient implementation of the Smith Waterman algorithm on the Nvidia GPU. The algorithm achieves more than 3.5 times higher per core performance than previously published implementation of the Smith Waterman algorithm on GPU, reaching more than 70% of theoretical hardware performance. The differences between current and earlier approaches are described showing the example for writing efficient code on GPU.

Efficacy of 2-halogen substituted d-glucose analogs in blocking glycolysis and killing “hypoxic tumor cells”

AbstractPurpose: Since 2-deoxy-D-glucose (2-DG) is currently in phase I clinical trials to selectively target slow-growing hypoxic tumor cells, 2-halogenated D-glucose analogs were synthesized for improved activity. Given the fact that 2-DG competes with D-glucose for binding to hexokinase, in silico modeling of molecular interactions between hexokinase I and these new analogs was used to determine whether binding energies correlate with biological effects, i.e. inhibition of glycolysis and subsequent toxicity in hypoxic tumor cells. Methods and Results: Using a QSAR-like approach along with a flexible docking strategy, it was determined that the binding affinities of the analogs to hexokinase I decrease as a function of increasing halogen size as follows: 2-fluoro-2-deoxy-D-glucose (2-FG) > 2-chloro-2-deoxy-D-glucose (2-CG) > 2-bromo-2-deoxy-D-glucose (2-BG). Furthermore, D-glucose was found to have the highest affinity followed by 2-FG and 2-DG, respectively. Similarly, flow cytometry and trypan blue exclusion assays showed that the efficacy of the halogenated analogs in preferentially inhibiting growth and killing hypoxic vs. aerobic cells increases as a function of their relative binding affinities. These results correlate with the inhibition of glycolysis as measured by lactate inhibition, i.e. ID50 1 mM for 2-FG, 6 mM for 2-CG and > 6 mM for 2-BG. Moreover, 2-FG was found to be more potent than 2-DG for both glycolytic inhibition and cytotoxicity. Conclusions: Overall, our in vitro results suggest that 2-FG is more potent than 2-DG in killing hypoxic tumor cells, and therefore may be more clinically effective when combined with standard chemotherapeutic protocols.

Boruta - A System for Feature Selection

Machine learning methods are often used to classify objects described by hundreds of attributes; in many applications of this kind a great fraction of attributes may be totally irrelevant to the classification problem. Even more, usually one cannot decide a priori which attributes are relevant. In this paper we present an improved version of the algorithm for identification of the full set of truly important variables in an information system. It is an extension of the random forest method which utilises the importance measure generated by the original algorithm. It compares, in the iterative fashion, the importances of original attributes with importances of their randomised copies. We analyse performance of the algorithm on several examples of synthetic data, as well as on a biologically important problem, namely on identification of the sequence motifs that are important for aptameric activity of short RNA sequences.

Adding numbers with DNA

A novel algorithm based on DNA computing for adding binary integer numbers is presented. It requires the unique representation of bits placed in test tubes treated as registers. Amplification step used for the carry operation allows one, in theory, to add numbers with the same quantity of elementary operations, regardless of the number of bits used for representation. New notation proposed in the paper allows for efficient and abstract description of the technical operations on DNA.

MODELING THE DNA-SOLVENT INTERFACE

We extend the technique of using perpendicular distribution functions to salt solutions around nucleic acids. Both solute density averaged and nonaveraged reference frames are considered and compared. Using a previous simulation of DNA in salt water of over a nanosecond in duration, the aqueous distribution functions were found to be well coveraged, whereas the salt perpendicular distribution functions were less well determined. Three-dimensional density reconstructions reliably showed the prominent solvation features with transferable functions. The number of solute atom types needed for reconstructions of a given precision was determined in the context of the reference simulation data set with the goal of achieving a required level of reconstruction quality.

A Rough Set-Based Model of HIV-1 Reverse Transcriptase Resistome

Reverse transcriptase (RT) is a viral enzyme crucial for HIV-1 replication. Currently, 12 drugs are targeted against the RT. The low fidelity of the RT-mediated transcription leads to the quick accumulation of drug-resistance mutations. The sequence-resistance relationship remains only partially understood. Using publicly available data collected from over 15 years of HIV proteome research, we have created a general and predictive rule-based model of HIV-1 resistance to eight RT inhibitors. Our rough set-based model considers changes in the physicochemical properties of a mutated sequence as compared to the wild-type strain. Thanks to the application of the Monte Carlo feature selection method, the model takes into account only the properties that significantly contribute to the resistance phenomenon. The obtained results show that drug-resistance is determined in more complex way than believed. We confirmed the importance of many resistance-associated sites, found some sites to be less relevant than formerly postulated and—more importantly—identified several previously neglected sites as potentially relevant. By mapping some of the newly discovered sites on the 3D structure of the RT, we were able to suggest possible molecular-mechanisms of drug-resistance. Importantly, our model has the ability to generalize predictions to the previously unseen cases. The study is an example of how computational biology methods can increase our understanding of the HIV-1 resistome.

All Relevant Feature Selection Methods and Applications

All-relevant feature selection is a relatively new sub-field in the domain of feature selection. The chapter is devoted to a short review of the field and presentation of the representative algorithm. The problem of all-relevant feature selection is first defined, then key algorithms are described. Finally the Boruta algorithm, under development at ICM, University of Warsaw, is explained in a greater detail and applied both to a collection of synthetic and real-world data sets. It is shown that algorithm is both sensitive and selective. The level of falsely discovered relevant variables is low—on average less than one falsely relevant variable is discovered for each set. The sensitivity of the algorithm is nearly 100 % for data sets for which classification is easy, but may be smaller for data sets for which classification is difficult, nevertheless, it is possible to increase the sensitivity of the algorithm at the cost of increased computational effort without adversely affecting the false discovery level. It is achieved by increasing the number of trees in the random forest algorithm that delivers the importance estimate in Boruta.

Musical Instruments in Random Forest

This paper describes automatic classification of predominant musical instrument in sound mixes, using random forests as classifiers. The description of sound parameterization applied and methodology of random forest classification are given in the paper. Additionally, the significance of sound parameters used as conditional attributes is investigated. The results show that almost all sound attributes are informative, and random forest technique yields much higher classification results than support vector machines, used in previous research on these data.

Active enhancer positions can be accurately predicted from chromatin marks and collective sequence motif data

BackgroundTranscriptional regulation in multi-cellular organisms is a complex process involving multiple modular regulatory elements for each gene. Building whole-genome models of transcriptional networks requires mapping all relevant enhancers and then linking them to target genes. Previous methods of enhancer identification based either on sequence information or on epigenetic marks have different limitations stemming from incompleteness of each of these datasets taken separately.ResultsIn this work we present a new approach for discovery of regulatory elements based on the combination of sequence motifs and epigenetic marks measured with ChIP-Seq. Our method uses supervised learning approaches to train a model describing the dependence of enhancer activity on sequence features and histone marks. Our results indicate that using combination of features provides superior results to previous approaches based on either one of the datasets. While histone modifications remain the dominant feature for accurate predictions, the models based on sequence motifs have advantages in their general applicability to different tissues. Additionally, we assess the relevance of different sequence motifs in prediction accuracy showing that even tissue-specific enhancer activity depends on multiple motifs.ConclusionsBased on our results, we conclude that it is worthwhile to include sequence motif data into computational approaches to active enhancer prediction and also that classifiers trained on a specific set ofenhancers can generalize with significant accuracy beyond the training set.