A bioinformatics expert system linking functional data to anatomical outcomes in limb regeneration

The current version of the formalism and database will be extended on several fronts. The main limitation of our current implementation is that it is constrained to two dimensions. Although the current formalism supports morphological segments reversed along a third dimension, the implementation of precise rotations will require the full support of three‐dimensional data. The mathematical representation based on graphs can be easily extended to handle three‐dimensional phenotypes, but will require implementation of new intuitive user interfaces for the visualization and manipulation of three‐dimensional morphologies. However, morphological data reported in the literature currently lack three‐dimensional information, due precisely to the lack of adequate tools for the processing and formalization of three‐dimensional phenotypes. Finally, the results stored in the database are currently limited to morphological outcomes; hence, the database does not include information regarding gene expression and protein patterns, genetic sequences, or microscopic tissue details from micrographs that may have been reported along with the experiments. Future versions of the formalism and database can gradually incorporate more and more detailed information and references to existing genetic databases as soon as the data‐driven models become robust enough to deal with fine‐grain morphological details.

Together, the presented formalism, centralized database, and software tool represent a powerful instrument for mining the extensive limb regeneration literature. The ever‐increasing amount of information disseminated throughout the scientific literature is currently encoded in natural language, photographs, and cartoon diagrams of diverse styles, which hinder true insight into mechanistic models of high‐level pattern regulation. Our formalism and database represent the first steps to systematically and unambiguously curate the current knowledge of limb regeneration, facilitating the development of mechanistic models explaining the full extent of the known results and improving the understanding of large‐scale patterning of complex structures. However, this is just the first step in our roadmap for a bioinformatics of shape.

Most importantly, the presented formalism and database is not only an indispensable tool for human scientists but, due to its mathematical nature, can be accessed and interpreted by automated algorithmic methods. Already, the data on limb regeneration are so vast that it is practically impossible to simply think up mechanistic models consistent with all functional data. Assistance from bioinformatics tools will become indispensable in most developmental and regenerative studies. Indeed, regenerative biology will benefit enormously from the application of computational modeling and simulation techniques to find candidate models (Iber and Zeller 2012) using the presented formalism and database of experiments. A complete formalization of experimental knowledge in limb regeneration, our system presented here being a solid foundation, will pave the way to applying algorithms for the automatic search of scientific discoveries (King et al. 2009; Sparkes et al. 2010).

Figure 6 shows a use case diagram of the presented formalization and database, including both human scientists and high‐performance computers. Human scientists can use the database to curate, centralize, and mine the repository of published experiments. In addition, high‐performance computers could simulate mechanistic models of regeneration and test their validity against the experiments stored in the centralized database. We are currently working on artificial intelligence tools that can discover mechanistic models of pattern regeneration, simulate these models in virtual cells, and use the database to determine whether the models correctly predict and explain the body of known data in limb regeneration.

In summary, we have presented the beginnings of a roadmap for a new bioinformatics of shape for limb regeneration experiments applicable to many model organisms. Our system is thus a candidate for standardizing the work in this field to facilitate dissemination and comparative analyses of results. Our mathematical approach for encoding and storing experimental, manipulative, and phenotypical information not only will help scientists in the field to access this vast amount of knowledge, but will allow the development of novel bioinformatics tools to help discover testable, mechanistic models from functional perturbation data.

Currently, the centralized database contains the details of 827 different experiments manually curated from 89 publications from the scientific literature and comprising hundreds of manipulations and morphologies (Table 1). The curation process of the centralized database was performed manually with the help of the software tool Limbform. We selected the most important work on limb regeneration from an extensive set of organisms, including salamanders, frogs, crustaceans, insects, and arachnids. Starting with the publications with the highest number of citations, we curated those experiments with altered morphological outcomes. Our selection includes limb regeneration experiments under a variety of manipulative operations such as cuttings, graftings, irradiations, and pharmacological treatments.

We used the database engine SQLite version 3 (public domain) for implementing the database schema designed for the presented formalism (Fig. 5). SQLite is the most used embedded relational database management system and implements most of the Structured Query Language (SQL), facilitating the interoperability with most of the current database infrastructures. An advantage of SQLite is that the database is contained in a single file, which includes both the schema and the data. This facilitates the administration and sharing of databases, since the database file can be easily copied, downloaded from the web, or sent by e‐mail to a collaborator.

We implemented the software tool Limbform as a native standalone desktop application for the Microsoft Windows, Mac OS X, and Linux platforms. The tool can create, read, and write databases following the presented scheme for limb regenerative experiments. Limbform stores a database as a single computer file, facilitating the sharing of databases among the research community. The curated centralized database of limb experiments can be downloaded as a single file and be opened for analysis or extension with Limbform.