Jhoirene B. Clemente

PROJECTION Algorithm for Motif Finding on GPUs

Motif finding problem is one of the NP-complete problems in Computational Biology. Existing nondeterministic algorithms for motif finding do not guarantee the global optimality of results and are sensitive to initial parameters. To address this problem, the PROJECTION algorithm provides a good initial estimate that can be further refined using local optimization algorithms such as EM, MEME or Gibbs. For large enough input (600-1000 base pair per sequence) or for challenging motif finding problems, the PROJECTION algorithm may run in an inordinate amount of time. In this paper we present a parallel implementation of the PROJECTION algorithm in Graphics Processing Units (GPUs) using CUDA. We also list down several major issues we have encountered including performing space optimizations because of the GPU’s space limitations.

Advice Complexity of the Online Search Problem

The online search problem is a fundamental problem in finance. The numerous direct applications include searching for optimal prices for commodity trading and trading foreign currencies. In this paper, we analyze the advice complexity of this problem. In particular, we are interested in identifying the minimum amount of information needed in order to achieve a certain competitive ratio. We design an algorithm that reads b bits of advice and achieves a competitive ratio of \((M/m)^{1/(2^b+1)}\) where M and m are the maximum and minimum price in the input. We also give a matching lower bound. Furthermore, we compare the power of advice and randomization for this problem.

Building phylogenetic trees from frequent subgraph mining techniques on reaction hypergraphs

Huge advances in experimental techniques have resulted in increasing amounts of biological network data being made available. Different kinds metabolic networks are just some of these. Analyzing the network topology of these metabolic networks across taxa can uncover important biological information that is independent of other currently available biological information. This study explores topological similarities between metabolic networks of different taxa to build phylogenetic trees. Similarities between graphs were determined using frequent subgraph mining techniques. Phylogenetic trees were then built based on the frequent subgraphs among reaction hypergraphs of different taxa. Experimental results show phylogenetic trees that bear a striking resemblance to those obtained from sequence comparisons.

Some Improvements of Parallel Random Projection for Finding Planted (l,d)-Motifs

A challenging variant of the motif finding problem (MFP) is called planted (l,d)-motif finding. It is where the pattern has expected number of mismatches on each of its occurrence in the sequence. Difficulty arises when the planted pattern is longer and the expected number of mismatches is high. An algorithm (FMURP) which uses a random projection technique followed by local search algorithms is shown to work better with longer motifs with higher accuracy. A parallel algorithm for FMURP already exist in the literature. However, the parallel algorithm uses an excessive amount of memory. In this paper, we propose some improvements on the existing parallel algorithm. We also prove that the modified parallelization is equivalent to the sequential version of FMURP.