OntoEffect: An OntoUML-Based Ontology to Explain SARS-CoV-2

Variants’ Effects

Ruba Al Khalaf

1 a

, Anna Bernasconi

1 b

and Alberto Garc

ıa S.

2 c

Politecnico di Milano, Milan, Italy

Universitat Polit

ecnica de Val

encia, Valencia, Spain

Keywords:

Ontological Analysis, OntoUML, SARS-CoV-2, Variants Impacts.

Abstract:

The SARS-CoV-2 virus continuously accumulates genetic variation through mutations; mutations are the

virus’ way to achieve viral adaptation. Although the huge amount of information accumulated on the virus

during the COVID-19 pandemic, the knowledge that contributes to explaining and supporting the research

related to SARS-CoV-2 characteristics and evolution is not currently organized, nor systematized. Here, we

present OntoEffect, an ontology that captures and represents such information systematically. Speciﬁcally, we

aim to represent the dimensions of the virus and its mutations, discussing their impacts on the virus itself, as

well as on public health, prevention, and treatment protocols. Aiming to obtain ontological clarity in such a

complex domain, OntoEffect was built using OntoUML, an ontology-driven conceptual modeling language,

grounded on the Uniﬁed Foundational Ontology (UFO). In the highly specialized context of virology, we show

the powerful ability of ontological models to provide clear and precise explanations of a domain and allow its

shared understanding among stakeholders.

1 INTRODUCTION

The severe acute respiratory syndrome coronavirus 2

(SARS-CoV-2) was initially detected in China in De-

cember 2019, causing the coronavirus disease 2019

(COVID-19). Consequently, after the continuous

spreading of the virus across the globe, the World

Health Organization (WHO) declared a Public Health

Emergency of International Concern (PHEIC) on 30

January 2020. Subsequently, on 11 March 2020, the

WHO ofﬁcially classiﬁed the outbreak as a pandemic.

After more than three years into the pandemic,

speciﬁcally on 5 May 2023, the WHO announced that

the disease no longer met the criteria for a PHEIC due

to its established and ongoing nature. According to

the same announcement, this decision does not imply

that the pandemic itself has ended, but rather that the

global emergency it caused has subsided, at least for

the time being (World Health Organization, 2023).

The impact of the global COVID-19 pandemic has

affected all human lives worldwide on multiple levels.

Since its emergence, the SARS-CoV-2 virus has con-

https://orcid.org/0000-0002-5645-5886

https://orcid.org/0000-0001-8016-5750

https://orcid.org/0000-0001-5910-4363

tinued to evolve genetically with an expanding range

of variants. As it happens in other RNA viruses,

achieving genetic diversity is an essential aspect for

the survival and continuation of SARS-CoV-2; in-

deed, it brings viral survival, ﬁtness, and pathogenesis

(Lauring et al., 2013; Mattenberger et al., 2021). The

observed impacts are continuously present, and they

vary depending on the speciﬁc variant that spreads

at given times of the outbreak. For example, it was

widely known that the viral transmission increased

with the Alpha variant of SARS-CoV-2 virus (Tanaka

et al., 2021). Therefore, collecting information about

the impacts of genetic variations in SARS-CoV-2 in

a systematic way, structuring it, and drawing the re-

lations between these impacts is of great importance.

In this research, we aim to ﬁll this gap; speciﬁcally,

we draw the relationships between the virus and the

impacts of its mutations and variants formation over

the progression of the pandemic.

In the context of COVID-19, it has been recog-

nized that many ontologies are being used to ad-

dress the challenges related to the pandemic analyt-

ics – with different intents like structural representa-

tion and semantics of data, visualizing the status of

infections, querying, and ontology editing and extrac-

tion (Ahmad et al., 2021).

Al Khalaf, R., Bernasconi, A. and S., A.

OntoEffect: An OntoUML-Based Ontology to Explain SARS-CoV-2 Variants’ Effects.

DOI: 10.5220/0012183900003598

In Proceedings of the 15th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2023) - Volume 2: KEOD, pages 62-72

ISBN: 978-989-758-671-2; ISSN: 2184-3228

Mapping knowledge into standardized models,

such as ontologies, allows for a precise understand-

ing of a particular domain (Romanenko et al., 2022;

Guizzardi and Guarino, 2023), which can be used to

support data representation, integration, sharing, and

analysis.

Along this direction, we present OntoEffect to for-

mally represent the set of relevant concepts in this

domain. More speciﬁcally, we capture the concepts

related to the impact of viral variation over the over-

all outcome of the virus behavior. OntoEffect also

captures the relationships between these concepts in

a structured format that can be used to integrate data

and facilitate knowledge sharing among researchers.

As methodological support for this process, we

employ OntoUML, a well-founded ontology-driven

conceptual modeling language whose meta-model

complies with the theoretically well-grounded Uni-

ﬁed Foundational Ontology (UFO) (Guizzardi, 2005).

OntoUML has previously been employed for on-

tological analysis of specialized life sciences do-

mains (Guizzardi et al., 2021; Garc

ıa S et al., 2022b)

and proven effective in improving the explanation of

complex domains with respect to traditional concep-

tual models (Garc

ıa S et al., 2022a; Verdonck et al.,

2019).

The scope of this research is broad, including

concepts that encompass -but are not restricted to-

epidemiology, immunology, and structural biology.

For achieving the highest possible clariﬁcation, we

follow the principles of the ontological unpacking

method (Bernasconi et al., 2022), proposed to analyze

and break down a complex concept or phenomenon

into its constituent parts based on an ontology that de-

ﬁnes the relationships between different types of en-

tities, ﬁnally resulting in a more granular understand-

ing of that complex system. In other words, the on-

tological unpacking method is a procedure in which

Ontology – as a ﬁeld of study – is utilized to apply

the theory of ontological analysis, thereby uncover-

ing the ontological conceptual model that depicts an

information structure.

The remainder of the paper is structured as fol-

lows. Section 2 reviews related works, including

well-known ontologies concerning the SARS-CoV-2

domain, and motivates our effort. Section 3 overviews

the methodology used to conduct the ontological anal-

ysis, listing the most relevant OntoUML class and re-

lationship stereotypes. Section 4 presents the Onto-

Effect ontology and its advantages. Finally, Section

5 brieﬂy describes the validation and concludes the

work.

2 RELATED WORK

Ontologies have gained attention during the pandemic

of COVID-19, and several researchers have built on-

tologies to solve diverse problems they were facing.

Within the list of UFO and OntoUML ontol-

ogy models presented in the OntoUML/UFO-catalog

(Barcelos et al., 2022) – a catalog for ontology-driven

conceptual modeling research (https://github.com/

unibz-core/ontouml-models), sharing all the models

that employ OntoUML for advancing ontology-driven

research – a pair of works related to what we are pre-

senting can be found:

• (Bernasconi et al., 2020) presented the Viral Con-

ceptual Model (VCM), which is a conceptual

model that focuses on viral genomic data struc-

turing to characterize genomic sequences.

• The same model went through an ontological un-

packing process (Guizzardi et al., 2021) achieved

with the use of the OntoUML language; the ob-

tained ontology represents the semantic areas of

Virus Infection, Tissue Sampiling, Virus Sequenc-

ing, and Virus Sequence Annotation. While the

VCM only served as a base for building a genomic

data structure, its ontological unpacking provided

a detailed analysis of the complex notions hidden

within the original VCM model.

• A domain ontology for modeling lockdown inter-

ventions during the COVID-19 pandemic was pre-

sented in (Fabio et al., 2021). It aims to provide a

comprehensive understanding of what a lockdown

entails by proposing an ontology-based modeling

approach. The authors highlight the need for a

well-grounded domain ontology that incorporates

real-world semantics to capture the complexities

and variations of lockdown measures.

Other related ontologies were developed with a dif-

ferent upper-level ontology, the Basic Formal Ontol-

ogy (BFO) (Arp et al., 2015). For instance, He et al.

presented the Coronavirus Infectious Disease Ontol-

ogy (CIDO, (He et al., 2020)) at the beginning of the

COVID-19 pandemic. CIDO is a community-based

open-source biomedical ontology that supports coro-

navirus disease knowledge, along with data standard-

ization, integration, sharing, and analysis. In addition

to being based on BFO, CIDO follows the Open Bi-

ological and Biomedical Ontologies (OBO) Foundry

Principles (Smith et al., 2007).

To our knowledge and after studying the domain,

we believe OntoEffect is unique in its scope and de-

sign approach as it systematically and comprehen-

sively describes the impact of viral variations on dif-

ferent levels and aspects.

OntoEffect: An OntoUML-Based Ontology to Explain SARS-CoV-2 Variants’ Effects

3 METHODOLOGY

The workﬂow for building OntoEffect is shown in

Figure 1. To identify the effects of interest that are

suited to be considered as entities in OntoEffect, we

ﬁrst built a taxonomy (i.e., controlled vocabulary) of

effects. In this article, we propose the transformation

of the taxonomy into a structured ontology.

We started from the beginning of the COVID-

19 pandemic, following a knowledge acquisition pro-

tocol to search the literature from different sources

considering both peer-reviewed articles and preprints,

with more focus on the peer-reviewed ones. Such

sources are Google Scholar, PubMed, GISAID/COG-

UK reports, MedRxiv, and BioRxiv. For keywords

search and to achieve relativeness to our ﬁeld of inter-

est, we used keywords composed of a virus-related

term (e.g., SARS-CoV-2), a known clinical impact

(e.g., infectivity), and a known mutation and/or vari-

ant (e.g., Alpha). The criteria for inclusion of an ef-

fect in the taxonomy of effect are its high representa-

tiveness in the literature and connection with at least

one non-synonymous mutation (or variant) of SARS-

CoV-2.

Along with the evolution of the virus, we con-

tinued curating the newest related publications in the

ﬁeld and adding new terms to the taxonomy. To struc-

ture the ﬁnal version of the taxonomy, we classiﬁed

the selected terms into four categories according to

the ﬁeld of impact.

They include Epidemiology (clustering the im-

pacts of a variant on epidemiological features); Im-

munology (impacts on the host immune system); Vi-

ral kinetics and dynamics (of viral proteins and the

viral cell cycle); and Diagnosis, prevention, and

treatments (impacts on these measures).

Note that different versions of the taxonomy were

proposed in past works: a preliminary idea is in

(Al Khalaf et al., 2021); consolidated in (Alfonsi

et al., 2022) (employed within a providing a complete

schema representing COVID-19 knowledge and data

interplay); ﬁnalized in (Serna Garc

ıa et al., 2023b),

used within the CoVEffect system, which mines the

effects of SARS-CoV-2 mutations and variants based

on deep learning (Serna Garc

ıa et al., 2023a).

Figure 1: Schematic workﬂow to build OntoEffect.

Lastly, after an informal exploration of the do-

main supported by experts in molecular biology and

bioinformatics, we started the ontological unpacking

process (Guizzardi et al., 2021) leading to identifying

units of our model (classes), starting from the previ-

ously developed terms’ deﬁnitions. Then, we also de-

ﬁned the relations among them. The transformation

process from each term in the taxonomy into OntoEf-

fect entities was performed manually and very care-

fully following the previously listed theories of On-

toUML, ﬁnally obtaining an ontological representa-

tion that reveals the ontological semantics of the built

taxonomy.

Here, we explored an alternative way to cluster

the effects of the taxonomy during the application of

the ontological unpacking process. We considered

what actor is impacted by each effect; this concep-

tual classiﬁcation allowed a clearer representation of

the relations between the different entities of the on-

tology. As a consequence, we obtained four clusters:

i) Viral-related, which contains the impacts related to

the virus characterization regardless of the host; ii)

Host-related, which contains the impacts related to

the host characterization regardless of the virus; iii)

Virus-host interaction, which contains the impacts re-

lated to the interactions between the virus and the host

of interest; and iv) Disease and disease management,

which contains the impacts involved after the disease

is established in the host and those related to the man-

agement of the disease (diagnosis, prevention, and

treatment protocols).

3.1 UFO and OntoUML Language

Foundational ontologies are comprehensive systems

of domain-independent categories and their connec-

tions based on solid philosophical principles. These

systems consist of axioms that deﬁne various el-

ements such as objects, events, causality, spatial-

temporal relationships, dependencies, and more.

They play a crucial role in representing phenom-

ena across different material domains. Some exam-

ples of foundational ontologies include Descriptive

Ontology for Linguistic and Cognitive Engineering

(DOLCE) (Borgo and Masolo, 2009), General Formal

Ontology (GFO) (Poli et al., 2010), Suggested Upper

Merged Ontology (SUMO) (Niles and Pease, 2001),

and Uniﬁed Foundational Ontology (UFO) (Guiz-

zardi, 2005). By providing a conceptual framework,

they enable us to grasp the fundamental nature of spe-

ciﬁc domains and make it easier to identify connec-

tions and interesting associations and/or relationships

among data that align with their ontological counter-

parts (Amaral et al., 2021).

According to (Verdonck and Gailly, 2016), UFO

and UFO-based conceptual modeling language (On-

KEOD 2023 - 15th International Conference on Knowledge Engineering and Ontology Development

toUML) have emerged among the most used ap-

proaches in the ﬁeld, and since the choice of an

upper-level ontology depends on the speciﬁc needs

and requirements of the application or domain be-

ing modeled. OntoUML is an ontology-driven con-

ceptual modeling language whose meta-model com-

plies with the ontological distinctions and axiomati-

zation of UFO (Guizzardi and Wagner, 2004). We

believe that UFO is the most suitable top-level ontol-

ogy to fulﬁll our needs of explanation of a complex

domain – as UFO informs OntoUML, which has been

used recently for this purpose in the life sciences do-

mains (Guizzardi et al., 2021; Bernasconi et al., 2022;

Garc

ıa S et al., 2022b). Therefore, we developed On-

toEffect using OntoUML language (Guizzardi, 2005).

OntoUML is an extension of the Uniﬁed Mod-

eling Language (UML) (Booch et al., 1997) based

on UFO. It incorporates ontological concepts and fo-

cuses on modeling not only software systems but also

the underlying conceptual structure and semantics of

the domains. OntoUML aims to represent entities, re-

lationships, and categories in a way that aligns with

formal ontology principles, enabling a more compre-

hensive understanding of complex systems by model-

ing them into ontologies. There are three theories in

OntoUML speciﬁcation listed as follows:

• Types and Individuals, OntoUML is founded

upon the essential distinction between types and

individuals. Types are abstract things that are

formed to assist in the perception and categoriza-

tion of the world around us. These things serve as

bundles of characteristics that can be anticipated

to be encountered in other speciﬁc entities - indi-

viduals which it can be endurants (e.g., objects)

or perdurants (e.g., events).

• Identity Principle, is a sort of function that is be-

ing used to differentiate two individuals, and ev-

ery individual must have exactly one identity prin-

ciple (e.g., a ﬁngerprint is an identity principle of

a person).

• Rigidity Principle, endurants can be either rigid

when an individual instantiates the same type in

all possible scenarios; or anti-rigid when it instan-

tiates a type only in a speciﬁc scenario.

Table 1 and Table 2 schematically represent the onto-

logical distinctions used in OntoEffect, both as class

stereotypes and relationships.

4 OntoEffect ARCHITECTURE

Here, we present the four views of OntoEffect,

corresponding to the listed clusters. The mas-

ter view that connects all these views (along with

the complete views) is available at https://tinyurl.

com/5n8c7k8t–not shown in the paper due to space

limitations. The schemata were drawn with Vi-

sual Paradigm Community Edition (v17.0) and the

OntoUML plugin, release 0.5.3 (https://github.com/

OntoUML/ontouml-vp-plugin). In the following sub-

sections, we wrote in bold-type the classes (i.e., enti-

ties) of the ontology and italic-type the relations be-

tween the entities.

4.1 Viral-Related View

We present a partition of the viral-related view in Fig-

ure 2. OntoEffect aims to represent the impact of viral

variation over the overall outcome of the virus behav-

ior; thus, the SARSCOV2 entity represents the virus

of our interest, which is already a mutated version

of the virus. We note that OntoEffect in its current

state does not include entities related to the formation

of a SARS-CoV-2 variant nor the composition of its

sequence. This is already modeled in the Virus Se-

quence Annotation view of the ontological unpacking

of the Viral Conceptual Model (VCM) (Bernasconi

et al., 2022; Guizzardi et al., 2021), which can be eas-

ily integrated with OntoEffect.

SARSCOV2 is an instance of VirusSpecies,

which itself is an instance of the Species.

Several SARSCOV2 can group forming

SARSCOV2Collective. A virus has a viral

membrane, a genome, and a proteome represented

as ViralMembrane, SARSCOV2Genome, and

SARSCOV2Proteome respectively.

A SARSCOV2Gene is a member of

SARSCOV2Genome, and it is linked through a

material relation with SARSCOV2Protein which

is a member of SARSCOV2Proteome. As an

example of SARSCOV2Protein, we present Spike

protein that is composed of two subunits S1 (which

contains the receptor binding domain, RBD) and

S2. Moreover, A viral protein could have in its

structure a region called Epitope that is a region

responsible for provoking an immune response in

the host. Each protein has its own characteristics,

such as Flexibility, Stability, Functioning, and

StructuralOptimization. Moreover, a protein might

include multiple NonSynonymousMutation.

Different types of Interactions could occur either

inside the same molecule (known as Intramolecular-

Interactions) or between molecules (known as In-

termolecularInteraction). Instead, IntraviralPro-

teinProteinInteractions is a bundle of relational dis-

positions connecting SARSCOV2Protein with itself

through an association.

OntoEffect: An OntoUML-Based Ontology to Explain SARS-CoV-2 Variants’ Effects

Table 1: The types of OntoUML class stereotypes used in OntoEffect.

Stereotype Deﬁnition

Type are types whose instances are themselves types.

Category is a rigid mixin that does not require a dependency to be speciﬁed. It is used to aggregate

essential properties of individuals who follow different identity principles.

Collective is used to represent rigid concepts that provide an identity principle for their instances.

The main characteristic of a hCollectivei is that it has a homogenous internal structure.

Kind is used to represent rigid concepts that provide an identity principle for their instances

and do not require a relational dependency. A hKindi represents a Functional Complex;

its parts contribute in different ways to its functionality.

Subkind is a construct used to represent rigid specializations of identity providers. By default, its

usage does not require a relational dependency.

Phase is used to represent anti-rigid subtypes of identity providers that are instantiated by

changes in intrinsic properties (e.g., the age of a person).

Role is a construct used to represent anti-rigid specializations of identity providers that are

instantiated in relational contexts. Like hPhasei , all instances of a hRolei must follow

the same identity principle.

Relator is used to represent truth-makers of material relations, i.e., the “things” that must exist

in order for two or more individuals to be connected by material relations.

Mode is a particular type of intrinsic property that has no structured value. Modes are individ-

uals that existentially depend on their bearers.

Quality is a particular type of intrinsic property that has a structured value. Qualities are things

that are existentially dependent on the things they characterize.

Other characteristics of the virus are the Incuba-

tionPeriod, which is the interval between the moment

a host is infected and the onset of the disease, and the

ViralVirulence which is the competence of the infec-

tious agent to produce pathologic effects.

Finally, a virus could incubate in a speciﬁc Cul-

ture, which can be speciﬁed through CultureCon-

ditions (such as Temperature), CultureMediums,

and the employed CellType and CellLine. We rep-

resent this incubation through a ViralCulture relator

in which we can assess the ViralFitness of the virus

in a speciﬁc environment.

4.2 Host-Related View

In Figure 3, a partition of the host-related view is pre-

sented. A Human is an instance of the HostSpecies,

which itself is an instance of the Species. This schema

comprises two sections; the right one describes the

different possible phases a human can go through,

while the left describes the human’s molecules, cells,

and systems.

A Sample is taken from a Human and is mediated

through a Diagnosis relator that results in a TestRe-

sult which could be a Positive or Negative which

characterize if the Human is in Infected phase or

Notinfected phase, respectively. A Human is in the

Infected phase and plays a HumanHost role. This

phase could have different generalizations:

• Livinig or Deceased host.

• Treated (i.e., playing a treatedIndividual role)

or Untreated host.

• According to the host’s history, the infected host

could be FirstTimeInfected or ReInfected with

the virus.

The NotInfected host plays a HumanReceiver role,

that is the human who has the ability to get the viral

infection from another host. Therefore, a material re-

lation links the HumanReceiver to the HumanHost,

and both are mediated through a ViralTransmission

relator. The NotInfected host is either NeverInfected

with the virus or got the infection earlier and already

Recovered from it.

Also, a Human could receive a vaccine and enter

the Vaccinated phase playing a VaccinatedIndivid-

ual role, or could stay UnVaccinated.

Human is more complex species than a Virus;

hence, it is composed of different systems, e.g. Im-

muneSystem which involves WhiteBloodCells. A

subkind of the WhiteBloodCells is Lymphocytes

KEOD 2023 - 15th International Conference on Knowledge Engineering and Ontology Development

Figure 2: A partition of the viral related view.

Table 2: The types of relations used in OntoEffect.

Relation Deﬁnition

Material hMateriali relations have a mate-

rial structure on their own. The

relata of a material relation are

mediated by individuals that are

called relators.

Mediation A relation between a hRelatori

and the entities it connects. It is

a type of existential dependence

relation. It can be derived from

the relation between the relata

and the individuals that compose

the relator and that inhere in the

relata.

Characterization A relation between a bearer type

and its feature. A feature is the

intrinsic (inherent) moment of

its bearer type, and thus existen-

tially dependent on the bearer.

ComponentOf Is a parthood relation between

two complexes.

MemberOf Is a parthood relation between

a functional complex or a

hCollectivei (as a part) and a

hCollectivei (as a whole).

which play a role in the immune response for ﬁght-

ing infectious agent (Wheelock and Toy, 1973). The

two main types of lymphocytes are Tcells and Bcells.

A HumanGene is a member of HumanGenome,

which is found exclusively in the HumanCells. The

HumanGene is linked through a material relation

with HumanProtein which is a member of Human-

Proteome. Also, HumanGene and HumanProtein

are participants of the HUMANProteinCodingPro-

cess.

The HostInfectedCells are speciﬁcally the group

of HumanCells that are infected with the virus (or

have the ability to be infected). Each of these cells has

a membrane (represented as HostCellMembrane)

that encloses different organelles, apparatus, and pro-

teins (represented as InfectedCellProtein).

Since the HumanProtein is a big class of

molecules, we only represent:

• InfectedCellProteins.

• ImmuneSystemProteins, which are the proteins

of the immune system, e.g. Immunoglobulin and

Antibody.

• HumanEnzyme like the Angiotensin-converting

enzyme 2 (ACE2) which plays a role as a Recep-

tor of the virus. Other enzymes are the Human-

Protease, like TMPRSS2 (i.e., Transmembrane

protease serine 2, playing a role as Transmem-

braneProtease), Furin (playing a role as Pro-

ProteinConvertase), and Cathepsin L.

Finally, we have the Neutralization category, which

involves different types of NeutralizingAgents that

can neutralize the virus and prevent it from enter-

ing the host cell. Such agents (besides the Anti-

bodies that are naturally part of the ImmuneSystem)

are TherapeuticAntibody as the laboratory manu-

factured antibodies, ConvalescentSera as the blood

serum that is obtained from an individual who has Re-

covered from a previous SARSCOV2 infection, and

VaccinatedSera as the blood serum that is obtained

from an individual who has been Vaccinated with one

of the SARS-CoV-2 vaccines. Finally, measurements

of the sensitivity of the virus to each of these agents

can be characterized as qualities of SensitivityToAn-

tibodies, SensitivityToConvalescentSera, and Sen-

sitivityToVaccinatedSera.

OntoEffect: An OntoUML-Based Ontology to Explain SARS-CoV-2 Variants’ Effects

Figure 3: A partition of the host-related view.

4.3 Virus-Host Interaction View

In Figure 4, a partition of the virus-host interac-

tion view is presented. To establish a viral infec-

tion, there are plenty of interactions between the virus

and its host that are needed. We collected these in-

teractions in one category that is called HostVirus-

Interaction, which is an instance of Interacrions.

ViralInfectionFormation, is a relator that mediates

SARSCOV2Collective and Human. To say that

a cell is infected with a virus, three main steps

must have happened (represented as instances of the

HostVirusInteraction):

1. ViralEntry, a category that represents the pro-

cess of a virus entering a host cell through one of

two entry pathways; CellSurfaceEntry and En-

dosomalEntry. The efﬁciency of this process can

be measured, and it is represented with a quality

called EntryEfﬁciency.

2. ViralReplication, after that the virus enters the

cell or its genome enters the host cell (through

EndosomalEntry or CellSurfaceEntry, respec-

tively). The SARSCOV2genome will be repli-

cated and translated into SARSCOV2Protein

through SARSCOV2ProteinCodingProcess

with the help of InfectedCellProtein.

3. ViralRelease, after that a host cell becomes in-

fected with the virus and starts to replicate the

virus and biosynthesis it, this cell becomes a

virus-producer cell which releases the mature

virus to infect other host cells in a process called

ViralRelease.

We are focusing on the viral entry process, as it is the

initial step of the ViralInfectionFormation. Prevent-

ing this process, either by a vaccine or a treatment,

means preventing the disease. Both entry pathways

start with the virus binding to the host cell receptor

in which the RBD of the Spike protein binds to the

ACE2 receptor. This process is presented through

the BindingToHostReceptorInteraction relator (has

BindingAfﬁnityToHostReceptor as a quality). This

binding induces conformational changes in the Spike,

which is followed by the S2 cleavage through Cleav-

ingInteraction. Depending on the entry route, the S2

is cleaved by different HumanProtease; using TM-

PRSS2 in the CellSurfaceEntry or using Cathepsin

L in the other pathway. Regardless of which entry

pathway the virus follows, a membrane fusion must

occur that is represented through MembraneFusion-

Interaction relator that mediates ViralMembrane

KEOD 2023 - 15th International Conference on Knowledge Engineering and Ontology Development

Figure 4: A partition of the Host-virus interaction view.

and HostCellMembrane since the fusion of these

two membranes is crucial to uncoating the genome

of SARS-CoV-2 to start the ViralReplication.

Just like MERS-CoV, the spike protein of SARS-

CoV-2 virus is cleaved by Furin into S1 and S2 sub-

units during their biosynthesis; therefore, this cleav-

age is also mediated through the CleavingInterac-

tion.

Another class of HostVirusInteraction repre-

sents the ImmuneReponse of the host toward the

presence of the virus inside it. To have such a re-

sponse, Epitope from a viral protein must be rec-

ognized by an Antibody or Immunoglobulin. The

immune response is a complex cascade of immune

mechanisms recruiting various immune cells and

chemical mediators. Therefore, we simpliﬁed it into

one relator called ImmuneResponse. A virus could

obtain the power to escape the host immune system

and minimize or completely vanish the ImmuneRe-

sponse; this phenomenon can be measured and is rep-

resented as ImmuneEscape. The binding between

Epitope and Antibody is mediated through Bind-

ingToAntibody relator (has BindingAfﬁnityToAnti-

body as a quality). Moreover, there is Neutraliza-

tionProcess relator that mediates the neutralization of

SARSCOV2 inside a HumanHost by the previously

mentioned NeutralizingAgent.

Since BindingToHostReceptorInteraction,

CleavingInteraction, NeutralizationProcess, and

BindingToAntibody are interactions between a

HumanProtein and a SARSCOV2Protein, we

generalized them in one category; ProteinProteinIn-

teraction.

4.4 Disease and Disease Management

View

In Figure 5, we present a partition of the view of On-

toEffect that characterizes the disease and its manage-

ment through diagnosis, treatment, and prevention.

We started this view form Human that could admit

into a Hospital (for many reasons, e.g., suspicion of

infection, receiving treatment or vaccine, or having

severe symptoms that require hospitalization) through

HospitalAdmission relator; also, many factors could

increase or decrease the RiskOfHospitalization of a

Human in a Hospital.

COVID19 is the Disease caused by SARSCOV2

infection and beside it an Infected host could suf-

fer from (or have) OtherDisease. Such a scenario

might lead to complications between COVID19 and

that OtherDisease and worsen the overall health sta-

tus of the Infected host; this was mediated through a

OntoEffect: An OntoUML-Based Ontology to Explain SARS-CoV-2 Variants’ Effects

Complications relator. The FatalityRate is a quality

that is calculated from the InfectedCollective and is

the proportion of persons who Deceased after the vi-

ral infection over the number of conﬁrmed Infected

people.

COVID19, as any other viral disease, is charac-

terized by InfectionDuration, which is the duration

of the host being infected with the virus. This disease

has different ranges of symptoms, which can be rep-

resented as phases which are: WithoutSymptoms,

MildSymptoms, ModerateSymptoms, and Sever-

Symptoms. The symptoms of the disease determine

its severity; therefore, the different phases of symp-

toms are associated with the DiseaseSeverity rela-

tor. Furthermore, an InfectedSample can be obtained

from the HumanHost to measure the ViralLoad in-

side it; ViralLoad is the number of copies of RNA of

a given virus per milliliter.

Lastly, there is a risk of a second infection, repre-

sented as Reinfection relator, by the same virus after

being Recovered from (or during the course of) a pri-

mary infection; such risk can be measured, and it is

represented with the RiskOfReinfection quality.

For disease diagnosis, as previously mentioned,

a Sample is taken from a Human and mediated

with Diagnostics collective through a Diagnosis re-

lator. A sample is used in a diagnostic test to de-

tect either the presence of SARSCOV2Protein (with

AntigenicTest, or MolecularTestPCR which has a

CtValue) or the presence of ImmuneSystemPro-

tein (with SerologicalTest) through DetectionPro-

cess. Such Diagnostics have measurements for their

effectiveness in detecting the presence of its target

molecules; we characterized it with a quality called

EffectivnessOfDiagnostics.

Considering the treatment of the disease, a cate-

gory of AntiviralDrug is mediated with the Treat-

ment relator and associated with different kinds

of drugs such as EntryInhibitors, AttachmentIn-

hibitors, and PolymeraseInhibitors. Another me-

diation relation with the Treatment relator is with

TreatedIndividual that is the role played by an

Treated host, and it represents an Infected host who

is Treated with any kind of AntiviralDrugs. The ef-

ﬁcacy of a drug to treat the infected host can be mea-

sured, and it is represented as EffectivnessOfDrug

quality.

Finally, considering the prevention protocols of

the disease, we are addressing the Vaccination rather

than other protocols (like self-isolation) since we are

focusing on the effects of the viral variation on the

outcome of the virus and its resulting disease. A Hu-

man enters the Vaccinated phase and plays a Vacci-

natedIndividual role if a Vaccine was received, and

the Vaccination relator mediates this process. A Vac-

cine is a NeutralizingAgent, and there are multiple

kinds of vaccines that were developed to prevent the

disease, e.g., Pﬁzer, Moderna, and AstraZeneca.

The efﬁcacy of a vaccine to neutralize the virus and

prevent viral infection formation can be measured,

and it is represented as EffectivnessOfVaccine qual-

ity.

5 VALIDATION AND

CONCLUSIONS

This paper presents OntoEffect, an OntoUML-based

ontology that explains SARS-CoV-2 variants’ effects.

As a core part of the research, we presented the On-

toUML model that represents classes and relation-

ships describing the domain. OntoEffect aims to pro-

vide a detailed analysis of the complex notions of the

impacts of SARS-CoV-2 variations. The main chal-

lenges of this effort are related to the complexity of

describing the biological domain with knowledge rep-

resentation artifacts. Many concepts are entangled or

nested, and their semantics are non-homogenous; thus

a deep understanding is needed and can be achieved

through an ontological representation.

Our current version of the ontology is undergo-

ing continuous updating of information. We designed

semi-supervised methods for extracting content from

the CORD-19 (Wang et al., 2020) literature cor-

pus to continuously collect instances of knowledge-

related entities (Serna Garc

ıa et al., 2023a). Un-

derstanding and mastering the complex domain of

SARS-CoV-2 mutation effects has proven extremely

useful, in our data-driven analysis of SARS-CoV-2

phenomena, where data science methods were in-

tegrated with knowledge-driven hypothesis making

(Bernasconi et al., 2021b; Bernasconi et al., 2021a).

As a preliminary assessment of this effort, we

presented the ontology to three domain experts, who

have a background in molecular biology and previ-

ously worked on several projects with viral genomic

sequences; they were asked to evaluate the ontology

for accuracy, relevance, and completeness. We had

two focus groups with the experts, resulting in a se-

ries of observations that will inspire our next reﬁne-

ment of the views. Generally, our modeling choices

were conﬁrmed (after we properly guided their un-

derstanding of OntoUML stereotypes). This work re-

quires broader validation by means of empirical stud-

ies, along the lines of (Verdonck et al., 2019; Garc

ıa S

et al., 2022a; Garc

ıa S. et al., 2023)–showing the

beneﬁts of this representation when the explanation

of such a complex domain is required to non-expert

KEOD 2023 - 15th International Conference on Knowledge Engineering and Ontology Development

Figure 5: A partition of the disease and disease management view.

stakeholders. All in all, we here defend that OntoEf-

fect presents a ﬁrst step towards the full understand-

ing of SARS-CoV-2 impacts to non-domain experts,

powering intra-domain knowledge exchange and in-

formation interoperability. This schema will further

enable knowledge-based applications for annotation

and reasoning.

REFERENCES

Ahmad, A., Bandara, M., Fahmideh, M., Proper, H. A.,

Guizzardi, G., and Soar, J. (2021). An overview of on-

tologies and tool support for COVID-19 analytics. In

2021 IEEE 25th International Enterprise Distributed

Object Computing Workshop (EDOCW), pages 1–8.

IEEE.

Al Khalaf, R., Alfonsi, T., Ceri, S., and Bernasconi, A.

(2021). Cov2k: a knowledge base of sars-cov-2 vari-

ant impacts. In Research Challenges in Information

Science: 15th International Conference, RCIS 2021,

Limassol, Cyprus, May 11–14, 2021, Proceedings,

pages 274–282. Springer.

Alfonsi, T., Al Khalaf, R., Ceri, S., and Bernasconi, A.

(2022). Cov2k model, a comprehensive representation

of sars-cov-2 knowledge and data interplay. Scientiﬁc

Data, 9(1):260.

Amaral, G., Baiao, F., and Guizzardi, G. (2021). Founda-

tional ontologies, ontology-driven conceptual model-

ing, and their multiple beneﬁts to data mining. Wiley

Interdisciplinary Reviews: Data Mining and Knowl-

edge Discovery, 11(4):e1408.

Arp, R., Smith, B., and Spear, A. D. (2015). Building on-

tologies with basic formal ontology. Mit Press.

Barcelos, P. P. F., Sales, T. P., Fumagalli, M., Fonseca,

C. M., Sousa, I. V., Romanenko, E., Kritz, J., and

Guizzardi, G. (2022). A fair model catalog for

ontology-driven conceptual modeling research. In

Conceptual Modeling: 41st International Conference,

ER 2022, Hyderabad, India, October 17–20, 2022,

Proceedings, pages 3–17. Springer.

Bernasconi, A., Canakoglu, A., Pinoli, P., and Ceri, S.

(2020). Empowering virus sequence research through

conceptual modeling. In Conceptual Modeling: 39th

International Conference, ER 2020, Vienna, Austria,

November 3–6, 2020, Proceedings 39, pages 388–

402. Springer.

Bernasconi, A., Guizzardi, G., Pastor, O., and Storey, V. C.

(2022). Semantic interoperability: ontological un-

packing of a viral conceptual model. BMC bioinfor-

matics, 23(Suppl 11):491.

Bernasconi, A., Gulino, A., Alfonsi, T., Canakoglu, A.,

Pinoli, P., Sandionigi, A., and Ceri, S. (2021a).

OntoEffect: An OntoUML-Based Ontology to Explain SARS-CoV-2 Variants’ Effects

VirusViz: comparative analysis and effective visual-

ization of viral nucleotide and amino acid variants.

Nucleic Acids Research, 49(15):e90–e90.

Bernasconi, A., Mari, L., Casagrandi, R., and Ceri, S.

(2021b). Data-driven analysis of amino acid change

dynamics timely reveals SARS-CoV-2 variant emer-

gence. Scientiﬁc Reports, 11(1):21068.

Booch, G., Rumbaugh, J., and Jackobson, I. (1997). The

Uniﬁed Modeling Language User Guide. Addison-

Wesley Object Technology Series – Pearson Educa-

tion (US).

Borgo, S. and Masolo, C. (2009). Handbook on Ontologies.

Springer Verlag.

Fabio, I., Amaral, G. C., Griffo, C., Bai

ao, F., and Guiz-

zardi, G. (2021). “What exactly is a lockdown?”: to-

wards an ontology-based modeling of lockdown inter-

ventions during the COVID-19 pandemic. In ONTO-

BRAS, pages 151–165.

Garc

ıa S, A., Bernasconi, A., Guizzardi, G., Pastor, O.,

Storey, V. C., and Costa, M. (2022a). An Initial Empir-

ical Assessment of an Ontological Model of the Hu-

man Genome. In Advances in Conceptual Modeling,

pages 55–65, Cham. Springer International Publish-

ing.

Garc

ıa S., A., Bernasconi, A., Guizzardi, G., Pastor, O.,

Storey, V. C., and Panach, I. (2023). Assessing the

value of ontologically unpacking a conceptual model

for human genomics. Information Systems, page

102242.

Garc

ıa S, A., Guizzardi, G., Pastor, O., Storey, V. C., and

Bernasconi, A. (2022b). An ontological characteri-

zation of a conceptual model of the human genome.

In Intelligent Information Systems: CAiSE Forum

2022, Leuven, Belgium, June 6–10, 2022, Proceed-

ings, pages 27–35. Springer.

Guizzardi, G. (2005). Ontological foundations for struc-

tural conceptual models. CTIT, Centre for Telematics

and Information Technology, Twente, Netherlands.

Guizzardi, G., Bernasconi, A., Pastor, O., and Storey, V. C.

(2021). Ontological unpacking as explanation: the

case of the viral conceptual model. In Conceptual

Modeling: 40th International Conference, ER 2021,

Virtual Event, October 18–21, 2021, Proceedings 40,

pages 356–366. Springer.

Guizzardi, G. and Guarino, N. (2023). Semantics, ontology

and explanation. arXiv preprint arXiv:2304.11124.

Guizzardi, G. and Wagner, G. (2004). A uniﬁed founda-

tional ontology and some applications of it in business

modeling. In Proceedings of the Open InterOp Work-

shop on Enterprise Modelling and Ontologies for In-

teroperability, co-located with CAiSE’04 Conference.

He, Y., Yu, H., Ong, E., Wang, Y., Liu, Y., Huffman, A.,

Huang, H.-h., Beverley, J., Hur, J., Yang, X., et al.

(2020). CIDO, a community-based ontology for coro-

navirus disease knowledge and data integration, shar-

ing, and analysis. Scientiﬁc data, 7:181.

Lauring, A. S., Frydman, J., and Andino, R. (2013). The

role of mutational robustness in RNA virus evolution.

Nature Reviews Microbiology, 11(5):327–336.

Mattenberger, F., Vila-Nistal, M., and Geller, R. (2021). In-

creased rna virus population diversity improves adapt-

ability. Scientiﬁc Reports, 11:6824.

Niles, I. and Pease, A. (2001). Towards a standard upper on-

tology. In Proceedings of the international conference

on Formal Ontology in Information Systems-Volume

2001, pages 2–9.

Poli, R., Healy, M., and Kameas, A. (2010). Theory

and applications of ontology: Computer applications.

Springer.

Romanenko, E., Calvanese, D., and Guizzardi, G. (2022).

Towards pragmatic explanations for domain ontolo-

gies. In Knowledge Engineering and Knowledge

Management: 23rd International Conference, EKAW

2022, Bolzano, Italy, September 26–29, 2022, Pro-

ceedings, pages 201–208. Springer.

Serna Garc

ıa, G., Al Khalaf, R., Invernici, F., Ceri, S., and

Bernasconi, A. (2023a). Coveffect: interactive sys-

tem for mining the effects of sars-cov-2 mutations

and variants based on deep learning. GigaScience,

12:giad036.

Serna Garc

ıa, G., Al Khalaf, R., Invernici, F., Ceri,

S., and Bernasconi, A. (2023b). Supporting

data for ”coveffect: Interactive system for min-

ing the effects of sars-cov-2 mutations and vari-

ants based on deep learning”, GigaScience Database.

http://dx.doi.org/10.5524/102386. Last accessed: Aug

2023.

Smith, B., Ashburner, M., Rosse, C., Bard, J., Bug, W.,

Ceusters, W., Goldberg, L. J., Eilbeck, K., Ireland, A.,

Mungall, C. J., et al. (2007). The OBO Foundry: coor-

dinated evolution of ontologies to support biomedical

data integration. Nature biotechnology, 25(11):1251–

1255.

Tanaka, H., Hirayama, A., Nagai, H., Shirai, C., Taka-

hashi, Y., Shinomiya, H., Taniguchi, C., and Ogata, T.

(2021). Increased transmissibility of the SARS-CoV-2

Alpha variant in a Japanese population. International

journal of environmental research and public health,

18(15):7752.

Verdonck, M. and Gailly, F. (2016). Insights on the use

and application of ontology and conceptual modeling

languages in ontology-driven conceptual modeling.

In Conceptual Modeling: 35th International Confer-

ence, ER 2016, Gifu, Japan, November 14-17, 2016,

Proceedings 35, pages 83–97. Springer.

Verdonck, M., Gailly, F., Pergl, R., Guizzardi, G., Mar-

tins, B., and Pastor, O. (2019). Comparing traditional

conceptual modeling with ontology-driven conceptual

modeling: An empirical study. Information Systems,

81:92–103.

Wang, L. L., Lo, K., Chandrasekhar, Y., Reas, R., Yang,

J., Eide, D., Funk, K., Kinney, R., Liu, Z., Merrill,

W., et al. (2020). CORD-19: The COVID-19 Open

Research Dataset. ArXiv.

Wheelock, E. F. and Toy, S. T. (1973). Participation of lym-

phocytes in viral infections. Advances in immunology,

16:123–184.

World Health Organization (2023). Statement on

the ﬁfteenth meeting of the IHR (2005) Emer-

gency Committee on the COVID-19 pan-

demic. https://www.who.int/news/item/05-05-

2023-statement-on-the-ﬁfteenth-meeting-of-the-

international-health-regulations. Last accessed: Aug

2023.

KEOD 2023 - 15th International Conference on Knowledge Engineering and Ontology Development