Processes 2018, 6,56
enhance the impression of inaccessibility and make reproduction challenging, even problematic.
Those characteristics can limit the credibility and acceptance of evidence and insights being presented
within computational biology reports. This overview illustrates specific ways in which methodological
and semantic clarity regarding mechanisms, explanations of phenomena and methods can be refined
to improve accessibility and strengthen methodological and scientific credibility.
In the context of mechanism-oriented models intended to better explain a biological phenomenon,
lack of clarity often involves the use of the terms “mechanistic” and “mechanistic model.” There is
considerable diversity in what is being implied when discussing mechanisms and/or describing
models as mechanistic. Mechanistic model is a convenient yet ambiguous phrase typically used
as an abbreviation for more accurate, more informative descriptors. Use of the term “mechanism”
is often similarly ambiguous. Clarity within research reports and credibility of claims made are
generally viewed as being correlated and computational biology is not an exception. Usage of
ambiguous phrases within research reports can limit the credibility and acceptance of the evidence
and insights being presented. This overview is motivated by ongoing collaborative efforts to improve
credibility (Supporting Material provides background) and the belief that improvements in semantic
and methodological clarity will strengthen the credibility of results leveraging simulation research.
The phrase “mechanistic model” has a variety of meanings ascribed to it that differ across
biological domains. There is an increasing tendency to utilize “mechanistic model” both specifically
and as an umbrella term. Herein, we define, characterize and cluster seven mechanistic model types
and suggest specific terms for each. To insure clarity, we narrow the scope of discussions that follow by
first limiting attention to reports seeking mechanism-oriented explanations of biological phenomena.
We further restrict focus to research for which a scientific objective is to (1) provide deeper, more
explanatory insight into the generation of biological phenomena; and/or (2) better predict, mimic, or
emulate one or more biological phenomena.
We clarify various uses of “mechanistic model” and how they are represented computationally
for explaining biological phenomena. We describe the spectrum of Mechanism-oriented Models and
methods being used to develop explanations of biological phenomena. We cluster explanations
of phenomena into three broad groups and then expand them into a total of seven model and
simulation types. We name each type and illustrate with diverse examples drawn from the literature.
We begin by framing the context and offering definitions. In “Methodological Complexity,” we
contrast how infrastructure and management of complexity influence clarity differently between
wet-lab and simulation research. In the section that follows, we describe three spectra that are useful
in describing, characterizing and distinguishing explanations of phenomena. Next, in “Three Groups
of Models of Explanation,” we use similarities and differences (with reference to the spectra) to
guide characterizations that distinguish semantically among seven workflow-centered models of
explanation, including four different types of computational models of explanation. The names used
to identify each characterization are not intended as semantic standards; rather they are offered
as suggestions to encourage movement in that direction and serve as a working foundation for an
ontology to use in explanatory simulation research in the life sciences. In “Relevant Information,
Multiple Sources,” we illustrate why providing sufficient methodological information is essential
to enhance the credibility of an explanatory simulation, whereas brevity weakens credibility at the
expense of clarity. We characterize five different sources and types of information from which relevant
details are needed to clearly distinguish among the four types of computational models of explanation.
In “Workflow, Provenance and Hybrid Models,” we comment on connections between workflows,
methods, and semantics and on new technical issues that further increase the need for semantic and
methodological clarity.
182
enhance the impression of inaccessibility and make reproduction challenging, even problematic.
Those characteristics can limit the credibility and acceptance of evidence and insights being presented
within computational biology reports. This overview illustrates specific ways in which methodological
and semantic clarity regarding mechanisms, explanations of phenomena and methods can be refined
to improve accessibility and strengthen methodological and scientific credibility.
In the context of mechanism-oriented models intended to better explain a biological phenomenon,
lack of clarity often involves the use of the terms “mechanistic” and “mechanistic model.” There is
considerable diversity in what is being implied when discussing mechanisms and/or describing
models as mechanistic. Mechanistic model is a convenient yet ambiguous phrase typically used
as an abbreviation for more accurate, more informative descriptors. Use of the term “mechanism”
is often similarly ambiguous. Clarity within research reports and credibility of claims made are
generally viewed as being correlated and computational biology is not an exception. Usage of
ambiguous phrases within research reports can limit the credibility and acceptance of the evidence
and insights being presented. This overview is motivated by ongoing collaborative efforts to improve
credibility (Supporting Material provides background) and the belief that improvements in semantic
and methodological clarity will strengthen the credibility of results leveraging simulation research.
The phrase “mechanistic model” has a variety of meanings ascribed to it that differ across
biological domains. There is an increasing tendency to utilize “mechanistic model” both specifically
and as an umbrella term. Herein, we define, characterize and cluster seven mechanistic model types
and suggest specific terms for each. To insure clarity, we narrow the scope of discussions that follow by
first limiting attention to reports seeking mechanism-oriented explanations of biological phenomena.
We further restrict focus to research for which a scientific objective is to (1) provide deeper, more
explanatory insight into the generation of biological phenomena; and/or (2) better predict, mimic, or
emulate one or more biological phenomena.
We clarify various uses of “mechanistic model” and how they are represented computationally
for explaining biological phenomena. We describe the spectrum of Mechanism-oriented Models and
methods being used to develop explanations of biological phenomena. We cluster explanations
of phenomena into three broad groups and then expand them into a total of seven model and
simulation types. We name each type and illustrate with diverse examples drawn from the literature.
We begin by framing the context and offering definitions. In “Methodological Complexity,” we
contrast how infrastructure and management of complexity influence clarity differently between
wet-lab and simulation research. In the section that follows, we describe three spectra that are useful
in describing, characterizing and distinguishing explanations of phenomena. Next, in “Three Groups
of Models of Explanation,” we use similarities and differences (with reference to the spectra) to
guide characterizations that distinguish semantically among seven workflow-centered models of
explanation, including four different types of computational models of explanation. The names used
to identify each characterization are not intended as semantic standards; rather they are offered
as suggestions to encourage movement in that direction and serve as a working foundation for an
ontology to use in explanatory simulation research in the life sciences. In “Relevant Information,
Multiple Sources,” we illustrate why providing sufficient methodological information is essential
to enhance the credibility of an explanatory simulation, whereas brevity weakens credibility at the
expense of clarity. We characterize five different sources and types of information from which relevant
details are needed to clearly distinguish among the four types of computational models of explanation.
In “Workflow, Provenance and Hybrid Models,” we comment on connections between workflows,
methods, and semantics and on new technical issues that further increase the need for semantic and
methodological clarity.
182