=Paper=
{{Paper
|id=Vol-3603/Paper1
|storemode=property
|title=Ontology-based Representation and Analysis of Conditional Vaccine Immune Responses
Using Omics Data
|pdfUrl=https://ceur-ws.org/Vol-3603/Paper1.pdf
|volume=Vol-3603
|authors=Anthony Huffman,Edison Ong,Tim Brunson,Nasim Sanati,Jie Zheng,Anna Maria Masci,Guanming Wu,Yongqun He
|dblpUrl=https://dblp.org/rec/conf/icbo/HuffmanOBS0MWH23
}}
==Ontology-based Representation and Analysis of Conditional Vaccine Immune Responses
Using Omics Data==
https://ceur-ws.org/Vol-3603/Paper1.pdf
Ontology-based representation and analysis of conditional
vaccine immune responses using Omics data
Anthony Huffman1, Edison Ong12, Tim Brunson3, Nasim Sanati3, Jie Zheng4, Anna Maria
Masci5, Guanming Wu3, Yongqun He6-8
1
Department of Computational Medicine and Bioinformatics, University of Michigan, Ann Arbor, MI, United
States.
2
GlaxoSmithKline, Rixensart 1330, Belgium.
3
Oregon Health & Science University, Portland, Oregon, OR, United States.
4
Department of Genetics, University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, United
States
5
Office of Data Science National Institute of Environmental Health Science, Durham, NC, United States.
6
Unit of Laboratory Animal Medicine, University of Michigan, Ann Arbor, MI, United States.
7
Department of Microbiology and Immunology, University of Michigan, Ann Arbor, MI, United States.
8
Center for Computational Medicine and Bioinformatics, University of Michigan, Ann Arbor, MI, United
States.
Abstract
ImmPort, the world’s largest repository of immunology data, includes many vaccine immune
response datasets. ImmPort maps the metadata of these studies to ontology and database
schema. As of February 28, 2023, our ImmPort data analysis identified 6.258 immune
exposures using 47 vaccines in 4,607 human subjects, and 324 cohort studies from the
ImmPort. We hypothesized that an integrative ontological representation of the data from these
studies would enhance our understanding and analysis of these ImmPort vaccine studies, and
with ontological classification and tools such as VIGET, we could further study the effects of
different conditions such as vaccine types and host biological sex on the vaccine response gene
expression profiles. Our Vaccine Ontology (VO) analysis classified these 37 vaccines into
bacterial, viral, and protozoan vaccine types with different vaccine properties. The ImmPort
metadata types were modeled with the Vaccine Investigation Ontology (VIO). Our new
ontology-based pipeline extracted vaccine response data from the ImmPort database, annotated
them based on ontology, obtained corresponding gene expression data from the GEO, and
performed consistent omics data analysis. Our use case found gene profiles shared and differed
from live and killed inactivated influenza vaccines. Furthermore, our Omics data analysis using
the VIGET tool found that female and male human subjects have differential host responses
for influenza vaccines. For example, our study showed much stronger early female responses
to influenza vaccination than males, and males was able to show active immune responses at a
later stage. Interestingly, the female (but not male) human subject group also showed
significantly enriched neutrophil degranulation at Day 3 after influenza vaccination; however,
males (but not females) displayed significantly enriched neutrophil degranulation at Day 14
after influenza vaccination. These mechanisms have been used to find differences between the
gene lists and pathways of host responses to different vaccines conditional to different factors
including vaccine types and host biological sex. Moreover, this framework can be expanded to
other vaccines and vaccine categories easily.
Keywords
Vaccine Ontology, ImmPort, GEO, Vaccine immune response, Gene expression profiles
Proceedings of the International Conference on Biomedical Ontologies 2023, August 28th-September 1st, 2023, Brasilia, Brazil
EMAIL: huffmaar@umich.edu (A. 1); edong@umich.edu (A. 2); tjbrunson@gmail.com (A. 3), nasim@plenary.org(A.4),
jiezhen@pennmedicine.upenn.edu(A.5), mascia2@nih.gov(A.6), wug@ohsu.edu(A.7) yongqunh@med.umich.edu (A.8)
ORCID: 0000-0002-5735-7540 (A. 1); 0000-0002-5159-414X (A. 2); 0000-0001-6681-0418 (A.4) 0000-0002-2999-0103 (A.5), 0000-0001-
9189-9661 (A.6), 0000-0001-8196-1177 (A.7), 0000-0001-9189-9661 (A.8)
© 2023 Copyright for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR
Wor
Pr
ks
hop
oceedi
ngs
ht
I
tp:
//
ceur
-
SSN1613-
ws.
or
0073
g
CEUR Workshop Proceedings (CEUR-WS.org)
CEUR
ceur-ws.org
Workshop ISSN 1613-0073
Proceedings
1
1. Introduction
As one of the greatest inventions in modern medicine, vaccination has been used to dramatically
protect humans against infectious diseases and improve human health. However, infectious diseases are
still a major cause of human mortality throughout the world, and effective vaccines that protect against
many diseases still do not exist [1,2]. The future success of effective vaccine development relies on
deep understanding of the protective immune mechanisms [3].
Vaccine induced host responses depend on various conditions and factors. For example, we have
previously developed the VaximmutorDB, a web-based database system that has included over 1,700
experimentally identified vaccine immune effectors (abbreviated as “vaximmutors”) induced by 154
vaccines for 46 pathogens [4]. The VaximmutorDB data have been manually annotated from peer-
reviewed publications and reliable databases. Our VaximmutorDB data analysis showed that these
vaccines induce many common immune factors, for example, Th1 immune factors IFN-gamma and IL-
2 and Th2 immune factors IL-4 and IL-6. Responses induced by different vaccine types such as
influenza and yellow fever vaccines may also differ [4]. However, more specific mechanisms
underlying vaximmutors and vaccine immune responses are still largely unclear, even when we
provided a successful use case for this yellow fever and influenzas vaccines for identifying these
vaximmutors [4]. Numerous other studies also identified many other factors crucial for vaccine response
induction [5-7]. For example, biological sex and age may change the immune responses significantly
[7, 8].
ImmPort (the Immunology Database and Analysis Portal; https://ImmPort.niaid.nih.gov/) is the
world’s largest repository of public-domain de-identified clinical trial data related to immunology [9].
All data derived from clinical trials funded by the Division of Allergy, Immunology and Transplantation
(DAIT) of the National Institute of Allergy and Infectious Diseases (NIAID) are required to be
published on the ImmPort portal. In addition, the ImmPort portal includes data obtained from the work
of the Human Immunology Project Consortium (HIPC, http://www.immuneprofiling.org/) as well as
relevant data from several external sources such as the Gates Foundation. ImmPort includes complete
clinical and mechanistic study data (e.g., mechanistic assays used, timing of visits, etc.), all of which
are publicly available for download in a de-identified form.
Pathway/network based analyses are extremely important in understanding protective immune
responses to infectious diseases. The immune systems of two vaccinees with different genetic
backgrounds may have different gene expression profiles in response to the same vaccine; however,
there may be shared underlying core immunological pathways between these two vaccinees. Various
experimental conditions may also affect host immune responses to vaccines. Nevertheless, appropriate
tools to support the analysis and visualization of vaccine-induced immune pathways under different
given conditions still miss.
We hypothesize that different vaccine types induce differential but coherent host responses, and
experimental conditions would significantly change the results of vaccine-induced host responses. More
specifically, we wanted to see if by leveraging ontology, that we could integrate multiple vaccine studies
to find common patterns for vaccines with the shared experiments. We have previously studied how
different factors would affect the host responses to Yellow fever vaccines [10], influenza vaccines [11],
and Brucella vaccines [12]. However, the previous studies were still limited with small datasets and
deep investigations. In current study, we would investigate further how specific factors such as
biological sex and vaccine types would change the vaccine-induced host responses.
Biomedical ontologies have emerged to be critical for the standardization, integration, and analysis
of the large amounts of heterogeneous biological data. We previously demonstrated how ontologies,
including the Vaccine Ontology (VO) [13], Vaccine Investigation Ontology (VIO) [14], and Ontology
of Biological and Clinical Statistics (OBCS) [3, 15] could be used to support vaccine induced host
response studies. Vaccine Ontology is a reference ontology focused on vaccines in general while
Vaccine Investigation Ontology is an extension that is focused on metadata types in various vaccine
investigation studies. Such approaches may be used to study the ImmPort data. Given the ImmPort
2
database with its own data management system, it would be important to integrate our ontology-based
approaches with the ImmPort data system for better studies.
VIGET is our newly developed web-based Vaccine response Gene Expression analysis Tool based
on Reactome and ImmPort [16]. VIGET uses the VO to classify various vaccine types and experimental
conditions. VIGET allows users to select vaccines with VO classification, choose ImmPort studies and
confounding variables, and perform differential gene expression analysis of various vaccine responses.
It is possible to apply VIGET to investigate the vaccine response gender differences by comparing
patterns for gender-based differences across influenza vaccines. We want to leverage our ontology
driven design to identify further patterns contained in multiple vaccine studies.
In this report, we demonstrated how ontology and VIGET can be applied to systematically
standardize, classify, integrate, and analyze vaccine-related high-throughput ImmPort and GEO data,
and identify vaccine-induced pathways under specific experimental conditions including vaccine types
and biological sexes of the human subjects. Three specific use cases were also developed to further
illustrate the effectiveness of our ontology-based representation and analysis of vaccine-induced host
responses given different vaccine or host conditions.
2. Methods
2.1. ImmPort vaccine metadata extraction and storage
The immune exposure metadata description of various vaccine studies was downloaded from the
public ImmPort website (https://www.immport.org/) to a local computer. Such metadata description is
stored in different ways in ImmPort. ImmPort provides the annotations to the gene expression data that
were deposited in GEO. The raw data can be downloaded from GEO. Our study downloaded the csv
text file of such metadata, and converted it to Excel for easy exploration and processing.
The ImmPort database has the immune exposure table that stores the information on how subjects
were immune exposed. Figure 1 shows how the vaccine immune exposure data in ImmPort was
standardized using the Vaccine Ontology (VO). From the data in ImmPort, we added additional
columns to create the metadata file via inter-rater discussion.
Figure 1: ImmPort immune exposure metadata illustration. The immune_exposure.txt file was
downloaded on March 28, 2023, from ImmPort website, opened in Excel format, and the vaccine
related information extracted. Note that it is only a portion of the contents in the file, and some
columns of the file are not shown here to save space. We added The Vaccine Ontology (VO) is used
here for vaccine information standardization.
3
2.2. Vaccine response data analysis using ontology-annotated ImmPort and
GEO
We used Ontofox [17] to generate a hierarchy of vaccine terms from VO that mapped to vaccine
terms studied in different experiments from ImmPort. We used the Vaccine Investigation Ontology
(VIO) to map and model metadata types for the vaccine experiments. Protégé-OWL editor was used for
ontology display and editing. SPARQL and DL-Query were used for ontology knowledge query.
2.3. Vaccine response data analysis using ontology-annotated ImmPort and
GEO
ImmPort provided vaccine information and the metadata for different vaccine response studies. The
vaccine types and metadata were annotated using VO and VIO. Based on the metadata of different
studies, we extracted the normalized microarray gene expression data from the NCBI GEO database
(https://www.ncbi.nlm.nih.gov/geo/). Software programs and APIs were generated to automatically
query the final integrative ontology and ImmPort data. As a use-case, influenza vaccine studies were
selected through ontology-based queries along with ontology-defined vaccine and condition (e.g.,
health and age) classification. For GEO data analysis, log2 transformation was applied if the expression
values of a given GEO dataset were not in log space. For each comparison pair, expression data
collected from individuals vaccinated on Day 0 was considered as the control group, while data
collected on Day 3 to Day 7 after vaccination were considered as the vaccine exposure group. In
addition, each pair was defined to have the same ontology-defined vaccine and condition within the
same cohort in the same ImmPort study.
2.4. Omics data analysis using VIGET
The web-based Reactome pathway analysis tool (https://reactome.org/) was used to support our gene
expression profile analyses. Adjusted FDR p-value cutoff of 0.05 was used for statistical data analysis
[18].
3. Results
3.1. Vaccine response data extraction from ImmPort
The ImmPort website maintains a publicly available data model schema
(https://immport.org/shared/dataModel) and a corresponding relational database. ImmPort does not
store raw Omics gene expression data. However, ImmPort provides the annotations to the gene
expression data that were deposited in GEO. The raw data can be downloaded from GEO.
An important table in ImmPort database is the immune exposure table that stores the information on
how subjects were immune exposed. Figure 1 shows how the vaccine immune exposure data in ImmPort
was standardized using the Vaccine Ontology (VO). Basically, the metadata file has many columns
including Exposure Accession, ARM Accession, Exposure Material ID (which uses VO IDs for
vaccines), Exposure Merial Reported (which is the label of the material), Exposure Process Reported
(which is “vaccination” for vaccine immune exposure studies), and Subject Accession. With the VO
IDs used in ImmPort metadata file, it became easy and efficient for us to standardize and process the
vaccine data and categorization.
Based on our analysis, as of February 28, 2023, the ImmPort database included 36,140 immune
records. Our analysis of the database identified 6,258 vaccine-related immune exposure records that
used 37 vaccine terms with unique VO IDs in 4,607 human subjects, 324 ARMs (a single group or
cohort for a specific study purpose).
4
3.2. VO-based classification of the ImmPort-reported vaccine studies
Figure 2: Ontological representation of 37 vaccines in ImmPort and their relations using VO. Vaccines
in VO are asserted by the pathogen they vaccinate against. Column A shows that ImmPort contains a
small selection of bacterial vaccines (7) and parasite vaccines (4). 3 of the 37 vaccines were mapped
to 2008-2009 influenza vaccine. The red boxes represent parent terms shown later in the hierarchy.
3.3. VIO-based of the vaccine investigation conditions
Figure 3 illustrates the basic flow to extract GEO expression data accession in the ImmPort database
schema by following six primary accessions. The flow chart represents the annotations under each GEO
GSM ID. Each study has one or more ARM (or cohort). Each ARM has one or more subjects. Each
subject has one or more biosample. Each biosample has one or more experimental sample (i.e.,
expsample). Meanwhile, each subject also has an immune exposure such as vaccination, and each
vaccination exposure points to a VO ID. Finally, each experimental sample corresponds to a specific
GSM ID. Each table also has more detailed annotations in the ImmPort database. Eventually, the GSM
ID can be used to get the data from the GEO database (Figure 3).
Figure 3: Vaccine response Omics data analysis using ontology-annotated ImmPort and GEO. Six IDs
in the boxes represent the primary keys of corresponding tables of the ImmPort database. These six
5
primary accessions are IDs for the study (study accession), cohort of a study (ARM accession), subject
of a cohort (subject accession), vaccine used on a subject (exposure accesion), biological test used on a
subject (biosample accession), the biological assay utilized on the subject (biosample accession), and
the experimental sample that was part of the biological assay (exsample accesion). The schema
illustrated here explains the components in Figure 1 and the relations among them, and links the
ImmPort data to GEO data.
3.4. Use case 1: Ontology-based query of vaccines and vaccine investigation
data
Based on the logically defined hierarchies and semantic relations in ontology, we can use our
ontology system to perform computer-assisted queries and analysis.
For example, based on the VO ontology hierarchy as shown in Figure 2, we can easily identify which
vaccines are bacterial vaccines, and which are viral vaccines. We can also detect which are influenza
vaccines or yellow fever vaccines. VO classification clearly lays out the relations of these vaccines.
Since the VO is computer-interpretable, the results can be automatically parsed by a semantic query
such as SPARQL or DL-query for various follow-up analyses or to aid retrieval of experimental studies
as part of use case 2.
Not shown in Figure 2, VO includes axioms that illustrate different properties of vaccines. For
example, the FluMist vaccine is defined to have live attenuated feature as seen in the following axiom:
‘has quality’ some ‘vaccine organism live attenuated’
In contrast, the Fluarix vaccine is defined to be an inactivated whole organism vaccine based on the
following axiom:
‘has quality’ some ‘vaccine organism inactivated’
Such logical axioms defined in VO provide us a logical way to automatically identify different
features. For example, we can use SPARQL or DL-query to easily find which vaccines are live
attenuated vaccines and which are inactivated vaccines. This was done to retrieve experimental studies
for further analysis of the vaccines within ImmPort.
Note that the ImmPort database does not provide the semantic information described above. We
extracted the VO IDs annotated in ImmPort and generated a VO subset that contains these VO IDs and
their corresponding vaccines and related vaccine attributes. This subset of VO was then applied to
support our semantic queries and analysis.
3.5. Use case 2: Detecting the effect of biological sex on gene expression
profiles stimulated by influenza vaccines
After we identify which vaccines or vaccine groups to evaluate, we can come back to query the
ImmPort database to obtain detailed vaccine study information (Figure 2). For example, using the
ontology strategy, we identified many vaccines grouped as inactivated influenza vaccines as shown in
Figure 2. Using ImmPort database, we identified three studies involving Fluzone and Fluarix, two
inactivated influenza vaccines which will be linked to GEO data as part of use case 2. It is known that
males and females may have different responses to vaccination. To see if using VIGET can recapitulate
this vaccine response gender difference, we compared patterns for gender-based differences across
influenza vaccines. We expanded our analysis to include non-immune related pathways related to cell
transcription or cell signaling. For influenza vaccines, we utilized all 16 influenza studies covering one
attenuated influenza vaccine, FluMist (VO_000044), and five inactivated influenza vaccine, Fluarix
(VO_000044), Fluzone (VO_000047), Fluvirin (VO_000046), the 2008-2009 trivalent inactivated
vaccine (VO_0004808) and the 2011-2012 trivalent inactivated vaccine (VO_0004810). These include
the same vaccines as part of use case 2. Our VIGET study included 210 male and 260 female subject
samples collected from either whole blood or peripheral blood mononuclear cells (PBMCs) [16]. These
6
subjects ranged from 0 years old to 90 years old and included the same ethnicities as the previous use
case. Finally, due to reduced sample sizes after 14 days, our analysis is focused on days 3, 7, and 14 in
comparison to day 0. All results were adjusted by age, race, and vaccine type. Using the VIGET tool,
we analyzed the effect on pathways using differences in log10-fold expression greater than 0.1, 0.2, 0.5,
1.0. These pathways can be found as part of Supplemental Table 1.
Our study showed that female vaccine responses tended to exhibit greater fold change than males
earlier to the time response. Using the default 0.2 pathway reveals that males exhibited less significant
pathways in females at all 3 time points (Table 1). Day 7 had the highest amount of significant genes
and pathways for both males and females. For males, significant pathways only emerged at log-fold
changes of 0.5 or greater; with the 1 log-fold expression showing neutrophil degranulation (FDR = 6.48
e-7) and innate immune system (FDR = 1.02e-2) as the only significant pathways. Females, in contrast,
had 10 pathways at the same time point (Table 1), including multiple immune response pathways,
including neutrophil degranulation (FDR = 5.06 e-11), IL-4 and IL-13 signaling (FDR = 2.31e-5), IL-
10 signaling (FDR = 4.01e-2), and the CLEC7A/inflammasome pathway (FDR = 4.77e-2).
Additionally, females at day 7 also exhibited cellular response to stress (FDR = 3.19e-4), with all genes
in this pathway being up-expressed (Supplemental Table 1, Flu-F-07-1.0-Pos.) Otherwise, there is a
consistent pattern of males having fewer significant pathways, and immune related pathways than
females for influenza vaccines.
Table 1. Summary of gender differences in number significant genes and pathways in influenza vaccine
response. Differential gene response for influenza vaccines across males and females. The initial
timepoint for comparison was Day 0 for all entries. All genes that had the magnitude of their log fold
change greater than the threshold were counted. For # of significant pathways, the numbers indicate
the number of pathways with FDR < 0.05. The full list of pathways can be found as part of the
Supplemental Table 1. Day 14 Males had no pathways related to immune response. All log-fold
changes are base 10. Results were corrected for age, race, batch, platform.
# of # of # of # of
Log-Fold Change
Time Comparison Significant Flu Significant Flu Significant Flu Significant Flu
Threshold
M Genes M Pathways F Genes F Pathways
0.1 176 17 969 73
0.2 3 35 56 126
Day 3
0.3 0 0 9 126
0.5 0 0 0 0
0.1 1388 0 1427 0
0.2 1329 0 1052 3
Day 7 0.3 1133 0 1048 0
0.5 1119 0 690 8
1.0 626 2 515 10
0.1 446 39 969 73
Day 14
0.2 44 36 98 70
7
0.3 4 68 11 34
0.5 0 0 0 0
Figure 4 shows the Reactome pathway enrichment analysis results at three days, seven days, and
fourteen days post-vaccination for males and females. Overall, females showed a significantly earlier
immune responses at Day 3 post influenza vaccination than males, and then later males caught up with
active immune responses. As shown in this figure, females exhibit an earlier vaccine response as shown
in pathways that are linked to signaling of interleukins 4, 13 and 10. An interesting finding is the
difference in neutrophil degranulation pathway expression in influenza-vaccinated female and male
groups. At Day 3 neutrophil degranulation was significantly enriched in the influenza-vaccinated
female group but not in the male group. At Day 7, both female and male groups showed significant
enrichment of neutrophil degranulation. However, in Day 14, only the male (but not female) group
showed significant enrichment of neutrophil degranulation (Figure 4).
Figure 4: Comparison of sex-based differences in immune response to influenza vaccines. All
Reactome subfigures were compared to Day 0 of vaccine administration. The conditions for subfigures
were labeled with text in the figure. Subfigures were generated by Reactome’s Reacfoam tool (Release
82, September 2022). The detailed information is provided in Supplemental Table 1.
We also investigated for sex differences that are common to live attenuated and inactivated influenza
vaccines. Due to a small data set for female inactivated influenza vaccine users, analysis for Day 0 to
Day 3 females could not be done using VIGET (Supplemental Table 2). However, inactivated influenza
8
vaccines revealed patterns of increased ribosomal translation as being enriched at Day 3 with fewer
immune pathways for males and females. Day 7 for the influenza vaccines failed to find pathways that
were found significant in males. Day 14 revealed 44 enriched immune related pathways. Males had
uniquely enriched pathways linked to the general immune system and interferon alpha/beta signaling
(Supplemental Table 2). Females, in contrast, had enriched pathways linked to innate immunity
enriched but not the general immune system. Intriguingly, day 7 and day 14 both reveal that females
have enriched pathways related to programmed cellular death and other cellular functions. Further
investigation of yellow fever vaccines revealed a similar pattern of additional enrichment of cell death
and cell signaling. Taken together, this shows that females have a common, unique response to the
vaccines we have tested.
4. Discussion
4.1. Application of Ontologies to ImmPort
The contributions of this study are multiple. First, we demonstrate our application of ontologies and
semantic relations to annotate the ImmPort data and link to GEO database. Second, we studied the effect
of vaccine type as a vaccine factor on the vaccine immune response profiles. Third, we detected the
effect of biological sex as an important host factor that affects the vaccine immune responses.
Our study shows that ontology can be applied with the relational ImmPort database to better support
vaccine immune response data analysis. There were initial difficulties with finding an appropriate
mapping between ImmPort and GEO due to the multiple lookup tables from each study to the list of
GEO IDs. While this was aided by ImmPort documentation, it still took significant effort trawling
through the database to find the matching IDs. There were also issues resolving if time matching
definitions for day 0 meant that data was taken before or after vaccination. However, once done, we can
now use this framework to expand to other vaccines or vaccine categories in ImmPort. As such, this
approach satisfies and aids in data FAIRness (Findable, Accessible, Interoperable, and Reusable). While
adaption of new axioms or ontology terms do require manual annotation, these can be easily done
following eXtensible Ontology Design (XOD) principles [19] and reusing a suite of Ontozoo tools
designed for this task [20, 21].Two database systems can be applied simultaneously. The ontology triple
store can be used to store ontology knowledge and metadata, and the relational database can be used to
store instance data. We can then use different query languages to link them together seamlessly. This
can be adapted to other databases, albeit after looking to find the best mapping for data. The use of
ontology can then be used to aid further intersections of different vaccination procedures (vaccine types,
vaccine timing, or different vaccination routes) and patient phenotypes (sex or age-based differences
between vaccine response). Ontology standardization has already been suggested to ImmPort to
standardize different cell types [22]. However, this was focused on mapping terms to biological entity
terms from Gene Ontology, Cell Ontology, Protein Ontology and the Ontology of Biomedical
Investigation. The use of ontology standardization for vaccines on these datasets is novel outside of
our lab.
It is to the best of our knowledge, that this is the first time that both databases were linked together
to gain information on the gene expression profile differences of TIV and LAIV. As GEO and ImmPort
have different focuses, this linkage will help facilitate greater understanding of the data. According to
a PubMed search using “GEO ImmPort database” (8/16/2021), GEO and ImmPort, have been used as
data sources to identifying genes related to cancer development and survival [23, 24].
4.2. The Effect of Biological Sex on Vaccine Immune Response for
ImmPort vaccines
Our study illustrates the significant effect of biological sex in the vaccine immune response
generation. The analysis of influenza vaccines showed females having an earlier and much stronger
9
activation of immune related pathways than males during the first week of vaccination. Moreover,
female immune responses at Day 7 uniquely had all genes part of cellular response to stress be
upregulated. It has been reported that females have a stronger early immune response than males to
vaccines [25], with females more likely to experience adverse events caused from an autoimmune
response, which may explain why these traits were only found when looking only at females. As this
pattern was found across live attenuated and dead influenza vaccines, this may be the result of vaccines
not being as optimized for human females and represents an avenue for further vaccine improvement
and mechanism research. It is interesting to find the significant differences in neutrophil degranulation
pathway expression in influenza-vaccinated female and male groups (Figure 6). The influenza-
vaccinated females induced fast neutrophil degranulation at the early stage (Day 3) than males. At Day
7, both female and male groups showed similar enrichment of neutrophil degranulation. However, only
males (but not females) had significant enrichment of neutrophil degranulation at Day 14 post
vaccination (Figure 6). Neutrophil degranulation is the regulated exocytosis of secretory granules
containing cellular mediators such as proteases and inflammatory mediators. Many recent studies infer
various roles of neutrophil degranulation in inflammatory disorder (e.g. septic shock and asthma) [26],
bacterial virulence strategy [27], and COVID-19 infection [28]. However, the role of neutrophil
degranulation in vaccine immune response induction is still unclear and worth further investigation.
Still, there are other studies that validate our results. A prior article that also identified increases of
ZAP-70 influenza immune response [29], suggesting that the role of ZAP-70 is duplicated in effective
vaccines and is part of gaining vaccine immunity. Another study has investigated differential gene
expression differences between older males and females from quadrivalent inactivated influenza virus
vaccines [7]. Quadrivalent influenza vaccines can be incorporated into this framework to help
understand how much of their differentially expressed genes are influenced by their vaccine type.
Readers can validate our results by going to the VIGET website (https://viget.violinet.org/).
In the future, we plan to further examine how ontology combined with tools such as VIGET [16]
can be used to further enhance our study of conditional vaccine immune responses given various
experimental and clinical conditions. For example, by linking the VaximmutorDB data with the
ImmPort data and Vaccine Ontology knowledge, we might be able to better understand the fundamental
mechanisms of protective vaccine immune responses. Different ontology-based machine learning
methods can also be explored to improve our prediction of new vaccine immune correlates and
mechanisms.
5. Conclusions
In summary, the development of this ontology-based framework is able to aid data standardization
and help guide further novel insights from the Omics data including the Omics metadata stored in the
ImmPort database and Omics raw data from GEO. These mechanisms have been used to find
differences between the pathways from different vaccines given different vaccine types and biological
sexes. Moreover, this framework can also be expanded to other vaccines and vaccine categories.
6. Acknowledgements
This project was funded by a NIH-NIAID UH2 grant (1UH2AI132931-01A1) and a NIH-NIAID
U24 grant (1U24AI171008). We appreciate the support provided by the ImmPort consortium. EO, YH
and GW contributed to the overall study design; JZ, AMM and YH provided manual verification and
ontology editing; EO, GW contributed to programming and data analysis; YH prepared the initial
manuscript, AH prepared the updated manuscript for the paper and all authors contributed to the
manuscript writing and reviews. All authors read and approved the final manuscript.
10
7. References
[1] Standing up to infectious disease. Nat Microbiol. 2019;4:1.
[2] Stutzer C, Richards SA, Ferreira M, Baron S, Maritz-Olivier C. Metazoan Parasite
Vaccines: Present Status and Future Prospects. Front Cell Infect Microbiol 2018, 8:67.
[3] Zhang C, Maruggi G, Shan H, Li J. Advances in mRNA Vaccines for Infectious
Diseases. Front Immunol 2019, 10:594.
[4] Berke K, Sun P, Ong E, Sanati N, Huffman A, Brunson T, Loney F, Ostrow J, Racz R,
Zhao B et al: VaximmutorDB: A Web-Based Vaccine Immune Factor Database and
Its Application for Understanding Vaccine-Induced Immune Mechanisms. Front
Immunol 2021, 12:639491.
[5] Beijnen EMS, Odumade OA, Haren SDV: Molecular Determinants of the Early Life
Immune Response to COVID-19 Infection and Immunization. Vaccines (Basel) 2023,
11(3).
[6] Zheng J, Li H, Liu Q, He Y: The Ontology of Biological and Clinical Statistics (OBCS)-
based statistical method standardization and meta-analysis of host responses to
yellow fever vaccines. Quant Biol 2017, 5(4):291-301.
[7] Yang J, Huang X, Zhang J, Fan R, Zhao W, Han T, Duan K, Li X, Zeng P, Deng J et al:
Sex-specific differences of humoral immunity and transcriptome diversification in
older adults vaccinated with inactivated quadrivalent influenza vaccines. Aging
(Albany NY) 2021, 13(7):9801-9819.
[8] Hilleman MR, Carlson AJ, Jr., McLean AA, Vella PP, Weibel RE, Woodhour AF:
Streptococcus pneumoniae polysaccharide vaccine: age and dose responses, safety,
persistence of antibody, revaccination, and simultaneous administration of
pneumococcal and influenza vaccines. Rev Infect Dis 1981, 3 Suppl:S31-42.
[9] Bhattacharya S, Dunn P, Thomas CG, Smith B, Schaefer H, Chen J, Hu Z, Zalocusky KA,
Shankar RD, Shen-Orr SS et al: ImmPort, toward repurposing of open access
immunological assay data for translational and clinical research. Scientific data 2018,
5:180015.
[10] Zheng J, Li H, Liu Q, He Y: The Ontology of Biological and Clinical Statistics (OBCS)-
based statistical method standardization and meta-analysis of host responses to
yellow fever vaccines. Quantitative Biology 2017, 5(4):291-301.
[11] Zheng J, Harris MR, Masci AM, Lin Y, Hero A, Smith B, He Y: OBCS: The Ontology
of Biological and Clinical Statistics. In: The 2014 International Conference on
Biomedical Ontologies (ICBO 2014): 2014; Houston, TX, USA. 2014: 1-6.
[12] Todd TE, Tibi O, Lin Y, Sayers S, Bronner DN, Xiang Z, He Y: Meta-analysis of
variables affecting mouse protection efficacy of whole organism Brucella vaccines
and vaccine candidates. BMC bioinformatics 2013, 14 Suppl 6:S3.
[13] Ozgur A, Xiang Z, Radev DR, He Y: Mining of vaccine-associated IFN-gamma gene
interaction networks using the Vaccine Ontology. Journal of biomedical semantics
2011, 2 Suppl 2:S8.
[14] Ong E, Sun P, Berke K, Zheng J, Wu G, He Y: VIO: ontology classification and study
of vaccine responses given various experimental and analytical conditions. BMC
Bioinformatics 2019, 20(Suppl 21):704.
11
[15] Zheng J, Harris MR, Masci AM, Lin Y, Hero A, Smith B, He Y: The Ontology of
Biological and Clinical Statistics (OBCS) for standardized and reproducible
statistical analysis. J Biomed Semantics 2016, 7(1):53.
[16] Brunson T, Sanati N, Huffman A, Masci AM, Zheng J, Cooke MF, Conley P, He Y, Wu
G: VIGET: A web portal for study of vaccine-induced host responses based on
Reactome pathways and ImmPort data. Front Immunol 2023, 14:1141030.
[17] Xiang Z, Courtot M, Brinkman RR, Ruttenberg A, He Y: OntoFox: web-based support
for ontology reuse. BMC research notes 2010, 3:175:1-12.
[18] Storey, J. D., & Tibshirani, R. (2003). Statistical significance for genomewide studies.
Proc. Natl. Acad. Sci. 2003, 100(16), 9440–9445.
[19] He Y, Xiang Z, Zheng J, Lin Y, Overton JA, Ong E: The eXtensible ontology
development (XOD) principles and tool implementation to support ontology
interoperability. J Biomed Semantics 2018, 9(1):3.
[20] Xiang Z, Courtot M, Brinkman RR, Ruttenberg A, He Y: OntoFox: web-based support
for ontology reuse. BMC Res Notes 2010, 3:175.
[21] Xiang Z, Zheng J, Lin Y, He Y: Ontorat: automatic generation of new ontology terms,
annotations, and axioms based on ontology design patterns. J Biomed Semantics 2015,
6:4.
[22] Overton JA, Vita R, Dunn P, Burel JG, Bukhari SAC, Cheung KH, Kleinstein SH, Diehl
AD, Peters B: Reporting and connecting cell type names and gating definitions
through ontologies. BMC Bioinformatics 2019, 20(Suppl 5):182.
[23] Wu C, Hu Q, Ma D: Development of an immune-related gene pairs signature for
predicting clinical outcome in lung adenocarcinoma. Sci Rep 2021, 11(1):3611
[24] Guo L, Wu Q, Ma Z, Yuan M, Zhao S: Identification of immune-related genes that
predict prognosis and risk of bladder cancer: bioinformatics analysis of TCGA
database. Aging (Albany NY) 2021, 13(15):19352-19374.
[25] Ciarambino T, Para O, Giordano M: Immune system and COVID-19 by sex differences
and age. Womens Health (Lond) 2021, 17:17455065211022262.
[26] Lacy P: Mechanisms of degranulation in neutrophils. Allergy Asthma Clin Immunol
2006, 2(3):98-108.
[27] Eichelberger KR, Goldman WE: Manipulating neutrophil degranulation as a bacterial
virulence strategy. PLoS Pathog 2020, 16(12):e1009054.
[28] Petito E, Franco L, Falcinelli E, Guglielmini G, Conti C, Vaudo G, Paliani U, Becattini C,
Mencacci A, Tondi F et al: COVID-19 infection-associated platelet and neutrophil
activation is blunted by previous anti-SARS-CoV-2 vaccination. Br J Haematol 2023.
[29] Nguyen THO, Sant S, Bird NL, Grant EJ, Clemens EB, Koutsakos M, Valkenburg SA,
Gras S, Lappas M, Jaworowski A et al: Perturbed CD8(+) T cell immunity across
universal influenza epitopes in the elderly. J Leukoc Biol 2018, 103(2):321-339.
12