In this poster we present the semantic and NLP layers in the development of our repository for experimental protocols. We have studied existing repositories for experimental protocols as well the experimental protocols themselves. We have identified end-‐user features across existing repositories; we have also structured the semantics for these documents, defined by an ontology and a Minimal Information model for experimental protocols. In addition, we have built an NLP layer that makes extensive use of semantics. Our integrative approach focuses on facilitating search, retrieval and socialization of experimental protocols. We also focus on facilitating the generation of documents that are born semantics.
Experimental protocols are fundamental information
structures that support the description of the processes by
means of which results are generated in experimental
research. Well-structured and accurately described protocols
(procesable by humans and machines) should facilitate
experimental reproducibility. In this poster we present the
semantic and NLP infrastructure that we are putting together
for machine procesable protocols; we emphasize in the
integration of key components of this infrastructure during
the implementation of a repository for experimental
protocols. Our components include: i) The SMART
Protocols (SP) Ontology: this ontology results from the
analysis of over 200 experimental protocols in various
domains –molecular biology, cell and developmental
biology and others. Domain experts also participated in the
development of the SP ontology
We have reviewed proposed standards for representing experimental protocols, investigations, experiments, scientific documents, rhetorical structures and annotations. In addition, we have analyzed existing repositories for protocols. Interestingly we have found that there are numerous similarities across these repositories –e.g. business model, end-user features, document management; by the same token, the lack of semantics for experimental protocols and the lack of specific features for this particular type of documents may be seen as a common deficiency in these repositories. This document is organized as follows; in section 2 the semantic components are presented; in this section we also inform on the use of semantics by our NLP engine. Some issues and final remarks are presented in section 3. 2
The combination of semantics and NLP makes it possible to deliver a tool that facilitates the generation of experimental protocols that are to be born semantics –fully annotated, linked to the web of data, with fully identified PoS, procesable by machines as well as by humans. In the same vein, a similar process for existing experimental protocols in formats such as PDF is also supported. Furthermore, searching for queries such as: “What bacteria have been used in protocols for persister cells isolation?”, “What imaging analysis software is used for quantitative analysis of locomotor movements, buccal pumping and cardiac activity on X. tropicalis?”, “How to prepare the stock solutions of the H2DCF and DHE dyes?”, is also possible.
We are using the SP ontology; SP aims to formalize the description of experimental protocols, which we understand as domain-specific workflows embedded within documents.
SP delivers a structured workflow, document and domain
knowledge representation written in OWL DL. For the
representation of document aspects we are extending the
1 http://gate.ac.uk/
Copyright c 2015 for this paper by its authors. Copying permitted for private and academic purposes
Information Artifact Ontology (IAO).2 The representation of
executable aspects of a protocol is captured with concepts
from P-Plan Ontology (P-Plan)
We are also working with the SIRO model; our model breaks down the protocol in key elements that are common to “all” laboratory protocols: i) Sample/Specimen (S), ii) Instruments (I), iii) Reagents (R) and iv) Objective (O).
SIRO is motivated by minimal information models as well as by the Patient/Population/Problem Intervention/Prognostic/Factor/Exposure Comparison Outcome (PICO) model. For the sample it is considered the strain, line or genotype, developmental stage, organism part, growth conditions, pre-treatment of the sample and, volume/mass of sample. For the instruments it is considered the commercial name, manufacturer and identification number. For the reagents it is considered the commercial name, manufacturer and identification number; it is also important to know the storage conditions for the reagents in the protocol. Identifying the objective or goal of the protocol, helps readers to make a decision about the suitability of the protocol for their experimental problem.
The four elements are also automatically annotated with existing ontologies and exposed as LOD.
The NLP engine, GATE, uses the semantics defined by the SP ontology and SIRO. We have classified our corpus of protocols according to purpose/objective (e.g. extraction of nucleic acids, DNA amplification and visualization of nucleic acids) and then we transformed them to text. For each protocol, metadata available, reagents, instruments samples, actions and instructions were manually identified.
We worked with full sentences to characterize PoS, relations, actions (verbs) and full instructions. Gazetteers and rules were thus generated. The results from our NLP workflow are very granular; for instance, we are able to identify DNA purification reagents, digest reaction reagents, cell disruption instruments, etc. Text like “plant species” is identified as sample, so are organisms and parts of organisms. The sentences and PoS where the vocabulary is located are also identified and characterized. For instance, PoS such as “leaf tissue finely ground using a mortar and pestle, then aliquoted (1 g) for each extraction” are 2 https://code.google.com/p/information-artifact-ontology/ 3 http://vocab.linkeddata.es/SMARTProtocols/sp-documentV2.0.htm 4 http://vocab.linkeddata.es/SMARTProtocols/sp-workflowV2.0.htm identified, characterized and annotated; in this example sample, action, cell disruption instrument are identified and characterized. We are using ANNIE (A Nearly-New Information Extraction) as our information extraction system and JAPE for coding rules.
We have presented the integration of three modules in the development of a repository for experimental protocols.
Unlike existing repositories, the SP repository focuses on facilitating the production of semantic protocols, intelligent search and retrieval and social activity over experimental protocols. We have extensively studied existing experimental protocols; key functionalities from these will also been included in our repository. We have also presented the SP ontology, the SIRO model for MI and the use of GATE in our architecture. Our workflow addresses scenarios with PDFs and de novo protocols – those born semantics based on the SP ontology. For de novo documents we are using the ontology as a template; the resulting instantiated RDF is annotated and the conventional document metadata is extracted. For PDFs we are tuning the NLP workflow for extracting SIRO automatically.
Extracting the Objective has proven to be a challenging task.
Actions e.g. grind the sample, usually have well defined grammatical structures; but, the Objective of the experimental protocol is usually hidden in a complex prose.
We are constantly improving the rules; new documents pertaining to other subdomains in biomedical sciences are added to the corpus; then, the rules are tested. Results are manually evaluated and the rules and gazetteers are consequently enriched. Copyright c 2015 for this paper by its authors. Copying permitted for private and academic purposes2