GRAPES: A Software for Parallel Searching on Biological Graphs Targeting Multi-Core Architectures

Abstract

Biological applications, from genomics to ecology, deal with graphs that represents the structure of interactions. Analyzing such data requires searching for subgraphs in collections of graphs. This task is computationally expensive. Even though multicore architectures, from commodity computers to more advanced symmetric multiprocessing (SMP), offer scalable computing power, currently published software implementations for indexing and graph matching are fundamentally sequential. As a consequence, such software implementations (i) do not fully exploit available parallel computing power and (ii) they do not scale with respect to the size of graphs in the database. We present GRAPES, software for parallel searching on databases of large biological graphs. GRAPES implements a parallel version of well-established graph searching algorithms, and introduces new strategies which naturally lead to a faster parallel searching system especially for large graphs. GRAPES decomposes graphs into subcomponents that can be efficiently searched in parallel. We show the performance of GRAPES on representative biological datasets containing antiviral chemical compounds, DNA, RNA, proteins, protein contact maps and protein interactions networks.

Portfolios of Subgraph Isomorphism Algorithms

“…In many applications, it is important to have the fastest possible algorithm even if the absolute differences in CPU time are small. For example, in pattern recognition [6,25] and chemical [8] applications, we often have to solve the subgraph isomorphism problem repeatedly for a very large number of graphs (in order to find a pattern image or molecule in a large database of target images or compounds, for example), so having an algorithm that is able to solve an instance in 100 ms instead of 1000 ms makes a big difference. Therefore, it is important to select the best algorithm for each instance, even if the instance is an easy one.…”

“…This NPcomplete problem has many important practical applications, for example in computer vision [6,25], biochemistry [8], and model checking [24]. There exist various exact algorithms, which have been compared on a large suite of instances by McCreesh and Prosser [15].…”

Performance and scalability of indexed subgraph query processing methods

“…We employ both real and synthetic graph datasets. Specifically, four real datasets of different characteristics are used: AIDS, PDBS, PCM, and PPI (see [9]). Furthermore, a very large number of synthetic datasets are generated that facilitate a systematic study on the dependence of the algorithms' performance and scalability on the key problem parameters (e.g., number of nodes, graph density, number of distinct labels, number of graphs, and query graph size).…”

“…These sets are not adequate to provide definitive conclusions on how an algorithm is influenced by the characteristics of the graphs. Of these works, Grapes [9] alone used several real datasets; however, the authors did not evaluate scalability. Also, their performance evaluation did not include a systematic exploration of the effect of the key problem parameters.…”

“…The iGraph comparison framework [10], which implements several such techniques, compared the performance of older algorithms (up to 2010). Since then, several, more efficient algorithms have been proposed (e.g., GraphGrepSX [2], Grapes [9], CT-Index [13]). Finally, virtually all of these works report on different metrics; thus, no concrete conclusion can be reached regarding their relative scalability.…”