A review of standards for data exchange within systems biology

Abstract

The rapid increase in experimental data within systems biology has increased the need for exchange of data to allow analysis and comparison of larger datasets. This has resulted in a need for standardized formats for representation of such results and currently many formats for representation of data have been developed or are under development. In this paper, we give an overview of the current state of available standards and ontologies within systems biology. We focus on XML-based standards for exchange of data and give a thorough description of similarities and differences of currently available formats. For each of these, we discuss how the important concepts such as substances, interactions, and experimental data can be represented. In particular, we note that the purpose of a standard is often visible in the structures it provides for the representation of data. A clear purpose is also crucial for the success of a standard. Moreover, we note that the development of representation formats is parallel to the development of ontologies and the recent trend is that representation formats make more and more use of available ontologies.

Improving the usability of standard schemas

“…This complexity problem is one that has also been shown to plague users in the field of biology. For example, Stromback et al [9] compare XML standard schemas for systems biology and found that one such standard schema, Sequentry XML, has a 26-level tree structure to represent the same information as three other systems biology XML standard schemas -BXML, INSDseq, and EMBxml -each of which have less than seven levels. Similarly, most of the legislative standard schemas introduced in [14] have complicated schemas that are composed of multiple sub-schemas.…”

“…However, a different Internet commerce XML standard schema such as Internet Open Trading Protocol (IOTP) [7] is required for the manufacturer to send the payment to the bank [3]. As in the domain of e-business, there is ample evidence for the existence of multiple standard schemas in the fields of finance, biology and legislation: between [27] and [8], nine different finance XML standard schemas were presented; in the field of systems biology [9] found eighty-five different XML standard schemas; and, [14] identified six different key legislative XML standard schemas in Europe.…”

“…As in the AEC domain, the e-business [4], finance [8], biology [9] [10] and legislation [12] [13] domains also commonly use standard schemas to address this heterogeneity at a syntactical level. In this section we validate that other domains have similar problems to those that we encountered in our case study, and in some cases these problems are exacerbated because the sheer quantity of data is significantly greater.…”

“…For example, in the finance domain, Knox [28] states that financial service providers are being pressured to decide which XML standard schemas to adopt, but implies that the choice is confusing due to so many competing and/or overlapping standard schemas. Stromback et al [9] state that even at a much smaller scale in the sub-domain of systems biology that looks at molecular interaction, there still exist major differences in the information represented by XML standard schemas [9]. In the ESTRELLA project, a comprehensive comparison of the available standard schemas for the legislation domain, it was necessary to extract "the best and most convincing principles" that could then be applied in developing a single integrated solution [14].…”

“…This process requires users, such as domain experts, to be able to understand multiple standard schemas and possibly compare them as well which exacerbates the problem. The frustration that domain practitioners have experienced with these problems has been well-documented in the evaluation of the use of standard schemas in practice in many individual domains (such as civil engineering [2], commerce [3,4,5,6], finance [7,8], biology [9,10,11] and legislation [12,13,14]). While these studies validate that standard schemas fall short in exactly the scenarios above, they are focused on fixing the standard schema for the domain rather than the more global, cross-domain problem of how to deal with the fact that standard schemas will inevitably not be able to be used for all needs in all applications.…”