Gabriela F. Minetti

Seeding strategies and recombination operators for solving the DNA fragment assembly problem

The fragment assembly problem consists in building the DNA sequence from several hundreds (or even, thousands) of fragments obtained by biologists in the laboratory. This is an important task in any genome project since the rest of the phases depend on the accuracy of the results of this stage. Therefore, accurate and efficient methods for handling this problem are needed. Genetic Algorithms (GAs) have been proposed to solve this problem in the past but a detailed analysis of their components is needed if we aim to create a GA capable of working in industrial applications. In this paper, we take a first step in this direction, and focus on two components of the GA: the initialization of the population and the recombination operator. We propose several alternatives for each one and analyze the behavior of the different variants. Results indicate that using a heuristically generated initial population and the Edge Recombination (ER) operator is the best approach for constructing accurate and efficient GAs to solve this problem.

Metaheuristic assemblers of DNA strands: Noiseless and noisy cases

The DNA fragment assembly problem is an NP-complete problem which has been solved efficiently by many metaheuristics. However, those techniques generally assemble fragments that belong to noiseless DNA sequences. But nowadays dealing with noisy instances is imperative. For that we analyse exhaustively how noiseless and noisy instances of this problem are dealt by three efficient algorithms (Problem aware local search, Simulated Annealing and Genetic Algorithims). This analysis includes a performance evaluation of those algorithms to assemble fragments and a study of the solution composition. From these analysis we observe that the GA is more robust in presence of noise than the other two searches, while it usually does not improve the accuracy of results for large instances (where Simulated Annealing is the more precise technique).

Variable size population in parallel evolutionary algorithms

Considering the population size is a critical parameter to define in evolutionary computation, in this paper an improved parallel evolutionary algorithm that incorporates different mechanisms to adapt the population size to the current status, is presented. Those mechanisms are based on resizing on fitness improvement GA (PRoFIGA) and variable population size (GAVaPS). Results indicate these incorporations are a reasonable choice when refinement in solutions is necessary.

An improved trajectory-based hybrid metaheuristic applied to the noisy DNA Fragment Assembly Problem

The DNA Fragment Assembly Problem (FAP) is an NP-complete that consists in reconstructing a DNA sequence from a set of fragments taken at random. The FAP has been successfully and efficiently solved through metaheuristics. But these methods usually face difficulties to succeed when noise appears in the input data or during the search, specially in large instances. In this regard, the design of more efficient techniques are indeed necessary. One example of these techniques found in literature is the Problem Aware Local Search (PALS) which represents a state-of-the-art and robust assembler to solve noisy instances. Although PALS performs better than other metaheuristics, the quality of the achieved solutions by this method can still be improved. Towards this aim, this work proposes a new hybrid and effective method that combines a local search technique specially designed for this problem (PALS) with Simulated Annealing (SA), which is further distributed following an island model. Our proposed hybrid approach showed an improved performance on the largest non-noisy and noisy instances when compared separately with Simulated Annealing and PALS.

Variable Neighborhood Search as Genetic Algorithm Operator for DNA Fragment Assembling Problem

Many specific algorithms and metaheuristics have been proposed for solving the DNA fragment assembly problem, but new algorithms with more capacity for solving this problem are necessary. The fragment assembly problem consists in building the DNA sequence from several hundreds (or even, thousands) of fragments obtained by biologists in the laboratory. This is an important task in any genome project since the rest of the phases depend on the accuracy of the results of this stage. In order to achieve this objective we propose a hybrid algorithm that achieves very accurate results in comparison with other metaheuristics.

The Problem Aware Local Search algorithm: an efficient technique for permutation-based problems

In this article, we will examine whether the Problem Aware Local Search, an efficient method initially proposed for the DNA Fragment Assembly Problem, can also be used in other application domains and with other optimization problems. The main idea is to maintain the key features of PALS and apply it to different permutation-based combinatorial problems. In order to carry out a comprehensive analysis, we use a wide benchmark of well-known problems with different kinds of variation operators and fitness functions, such as the Quadratic Assignment Problem, the Flow Shop Scheduling Problem, and the Multiple Knapsack Problem. We also discuss the main design alternatives for building an efficient and accurate version of PALS for these problems in a competitive manner. In general, the results show that PALS can achieve high-quality solutions for these problems and do it efficiently.

Developing Genetic Algorithms Using Different MapReduce Frameworks: MPI vs. Hadoop

MapReduce is a quite popular paradigm, which allows to no specialized users to use large parallel computational platforms in a transparent way. Hadoop is the most used implementation of this paradigm, and in fact, for a large amount of users the word Hadoop and MapReduce are interchangeable. But, there are other frameworks that implement this programming paradigm, such as MapReduce-MPI. Since, optimization techniques can be greatly beneficiary of this kind of data-intensive computing modeling, in this paper, we analyze the performance effect of developing genetic algorithms (GA) using different frameworks of MapReduce (MRGA). In particular, we implement MRGA using Hadoop and MR-MPI frameworks. We analyze and compare both implementations considering relevant aspects such as efficiency and scalability to solve a large dimension problem. The results show a similar efficiency level between the algorithms but Hadoop presents a better scalability.

Constrained two-dimensional non-guillotine cutting problem an evolutionary approach

General cutting problems are concerned with finding the best allocation of a number of items in larger containing regions. These problems can be encountered in numerous areas such as computer science, industrial engineering, logistics, manufacturing, among others. They belong to the family of NP-complete problems. For cases of high complexity deterministic and exact techniques become inefficient due to the vast number of possible solutions that have to be evaluated. In order to reduce the computational load, heuristic or meta-heuristic algorithms are used. The solution method presented in this paper is meta-heuristic based on an evolutionary approach, being its goal to maximize the total value of cut pieces. For that, a modification of Beasleys representation is adopted and for evaluating solutions three placement heuristic rules are developed. Moreover, the effect that placement rules has on evolutionary algorithms performance is tested. Computational results are presented for a number of test problems taken from the literature. The results are very encouraging.