Biocompute Objects-A Step towards Evaluation and Validation of Biomedical Scientific Computations

Vahan Simonyan; Jeremy Goecks; Raja Mazumder

doi:10.5731/pdajpst.2016.006734

Biocompute Objects-A Step towards Evaluation and Validation of Biomedical Scientific Computations

PDA J Pharm Sci Technol. 2017 Mar-Apr;71(2):136-146. doi: 10.5731/pdajpst.2016.006734. Epub 2016 Dec 14.

Authors

Vahan Simonyan¹, Jeremy Goecks², Raja Mazumder³

Affiliations

¹ Center for Biologics Evaluation and Research, Food and Drug Administration, Silver Spring, MD, USA; vahan.simonyan@fda.hhs.gov jgoecks@gwu.edu mazumder@gwu.edu.
² Computational Biology Institute, George Washington University, Ashburn, VA, USA; and vahan.simonyan@fda.hhs.gov jgoecks@gwu.edu mazumder@gwu.edu.
³ Department of Biochemistry and Molecular Medicine, George Washington University, Washington, DC, USA vahan.simonyan@fda.hhs.gov jgoecks@gwu.edu mazumder@gwu.edu.

Abstract

The unpredictability of actual physical, chemical, and biological experiments due to the multitude of environmental and procedural factors is well documented. What is systematically overlooked, however, is that computational biology algorithms are also affected by multiplicity of parameters and have no lesser volatility. The complexities of computation protocols and interpretation of outcomes is only a part of the challenge: There are also virtually no standardized and industry-accepted metadata schemas for reporting the computational objects that record the parameters used for computations together with the results of computations. Thus, it is often impossible to reproduce the results of a previously performed computation due to missing information on parameters, versions, arguments, conditions, and procedures of application launch. In this article we describe the concept of biocompute objects developed specifically to satisfy regulatory research needs for evaluation, validation, and verification of bioinformatics pipelines. We envision generalized versions of biocompute objects called biocompute templates that support a single class of analyses but can be adapted to meet unique needs. To make these templates widely usable, we outline a simple but powerful cross-platform implementation. We also discuss the reasoning and potential usability for such concept within the larger scientific community through the creation of a biocompute object database initially consisting of records relevant to the U.S. Food and Drug Administration. A biocompute object database record will be similar to a GenBank record in form; the difference being that instead of describing a sequence, the biocompute record will include information related to parameters, dependencies, usage, and other information related to specific computational instance. This mechanism will extend similar efforts and also serve as a collaborative ground to ensure interoperability between different platforms, industries, scientists, regulators, and other stakeholders interested in biocomputing.

Keywords: Biocompute object; Computation reproducibility; FDA; NGS standardization; Regulatory research.

MeSH terms

Computational Biology / legislation & jurisprudence
Computational Biology / methods*
Databases, Factual
High-Throughput Nucleotide Sequencing / instrumentation
High-Throughput Nucleotide Sequencing / methods*
Reproducibility of Results
Sensitivity and Specificity
United States

Abstract

MeSH terms

Grants and funding