Identifying frequent patterns in biochemical reaction networks: a workflow

Abstract

Computational models in biology encode molecular and cell biological processes. Many of these models can be represented as biochemical reaction networks. Studying such networks, one is mostly interested in systems that share similar reactions and mechanisms. Typical goals of an investigation thus include understanding of model parts, identification of reoccurring patterns and recognition of biologically relevant motifs. The large number and size of available models, however, require automated methods to support researchers in achieving their goals. Specifically for the problem of finding patterns in large networks only partial solutions exist. We propose a workflow that identifies frequent structural patterns in biochemical reaction networks encoded in the Systems Biology Markup Language. The workflow utilizes a subgraph mining algorithm to detect the network patterns. Once patterns are identified, the textual pattern description can automatically be converted into a graphical representation. Furthermore, information about the distribution of patterns among a selected set of models can be retrieved. The workflow was validated with 575 models from the curated branch of BioModels. In this paper, we highlight interesting and frequent structural patterns. Furthermore, we provide exemplary patterns that incorporate terms from the Systems Biology Ontology. Our workflow can be applied to a custom set of models or to models already existing in our graph database MaSyMoS. The occurrences of frequent patterns may give insight into the encoding of central biological processes, evaluate postulated biological motifs or serve as a similarity measure for models that share common structures.

Harmonizing semantic annotations for computational models in biology

“…Researchers cannot readily search across multiple repositories to find models of interest, and when a researcher does retrieve a relevant model, they must determine whether it (or part of it) can be repurposed for use in their modeling work. Similarity measures and pattern-matching algorithms can help identify relevant modeling components [ 28 , 49 ], but existing proposals for cross-repository search and retrieval [ 32 ] are mostly theoretical and not yet applicable in practice. Applying semantic annotations in a standardized way across model repositories would provide a common ground for such cross-repository searches, allowing models encoded in diverse formats to be retrieved based on their shared semantic descriptions.…”

SBGN Bricks Ontology as a tool to describe recurring concepts in molecular networks

“…In addition, extending BKO with sets of bricks expressed in other standard languages such as BioPAX will allow the development of new converters grounded in a formal framework. Such an extension will also open up the possibility to integrate the annotation process we used to analyze the ACSN and PANTHER databases into more systematic workflows [19] while using such model repositories as BioModels [4], Physiome Model Repository [28] or Reactome [10]. Formally represented building blocks of SBGN maps will also enable integration of graphical information into sophisticated model retrieval tools such as MaSyMoS [12].…”

Key challenges facing data-driven multicellular systems biology

“…Graph structures could also be applied to represent and visualize cell-cell interactions in multicellular populations [18], similarly to phylogenetic trees (e.g., [22, 23]), chemical reaction networks (e.g., [25, 26]), gene network diagrams (e.g., [27]), and emerging data formats for agent-based model rules (e.g., as in Morpheus [28]).…”