Diseasome and comorbidities complexities of SARS-CoV-2 infection with common malignant diseases

Abstract

With the increasing number of immunoinflammatory complexities, cancer patients have a higher risk of serious disease outcomes and mortality with SARS-CoV-2 infection which is still not clear. In this study, we aimed to identify infectome, diseasome and comorbidities between COVID-19 and cancer via comprehensive bioinformatics analysis to identify the synergistic severity of the cancer patient for SARS-CoV-2 infection. We utilized transcriptomic datasets of SARS-CoV-2 and different cancers from Gene Expression Omnibus and Array Express Database to develop a bioinformatics pipeline and software tools to analyze a large set of transcriptomic data and identify the pathobiological relationships between the disease conditions. Our bioinformatics approach revealed commonly dysregulated genes (MARCO, VCAN, ACTB, LGALS1, HMOX1, TIMP1, OAS2, GAPDH, MSH3, FN1, NPC2, JUND, CHI3L1, GPNMB, SYTL2, CASP1, S100A8, MYO10, IGFBP3, APCDD1, COL6A3, FABP5, PRDX3, CLEC1B, DDIT4, CXCL10 and CXCL8), common gene ontology (GO), molecular pathways between SARS-CoV-2 infections and cancers. This work also shows the synergistic complexities of SARS-CoV-2 infections for cancer patients through the gene set enrichment and semantic similarity. These results highlighted the immune systems, cell activation and cytokine production GO pathways that were observed in SARS-CoV-2 infections as well as breast, lungs, colon, kidney and thyroid cancers. This work also revealed ribosome biogenesis, wnt signaling pathway, ribosome, chemokine and cytokine pathways that are commonly deregulated in cancers and COVID-19. Thus, our bioinformatics approach and tools revealed interconnections in terms of significant genes, GO, pathways between SARS-CoV-2 infections and malignant tumors.

Introduction

Coronavirus disease-19 (COVID-19) caused by the SARS-CoV-2 virus has become a global crisis where the World Health Organization (WHO) declared it as a pandemic on 11 March 2020 [1]. This virus initially creates a respiratory illness that can spread rapidly. In addition to losing thousands of human lives, COVID-19 causes massive damages in the global economy. When numerous coronaviruses were studied, only seven are known that affects human health and severe diseases have happened for three of them, including severe acute respiratory syndrome coronavirus (SARS-CoV), middle east respiratory syndrome coronavirus (MERS-CoV) and, the current pandemic, SARS-CoV-2 virus [2]. For SARS-CoV and MERS-CoV, two serious global epidemics happened in 2003 and 2012 [3], respectively, but did not declare them as a pandemic. However, SARS-CoV-2 is a single-stranded RNA virus that showed 89.1% nucleotide similarity and spread more easily than others. COVID-19 patients with a number of pre-existing medical conditions (e.g., diabetes, heart disease, cancer) are more likely to suffer severe COVID-19 and poor therapeutic outcomes compared to normal infected people. Indeed, this virus affects multiple organs severely in the human body. Regarding cancer patients, a study was conducted over 55,000 confirmed COVID-19 cases in China where the death rate was 7.6% that indicated five times higher death risk than COVID-19 patients without comorbidities (1.4%) [4].Due to the relative weakness of patients for COVID-19, the question has been risen about the effects of various cancers and associated comorbidities. There is no adequate evidence about direct interaction among COVID-19 and various cancers. The frailty and cancer therapeutics are not easily modifiable where the interactions happened due to the cellular pathways of cancers and SARS-CoV-2 that could be focused by therapeutic intervention.

Numerous works of COVID-19 and cancer gene expressions happened to investigate and identify altered pathways that could serve as resources for studying COVID-19 and its cancer comorbidities. Also, it causes the changes of many potentially shared molecular factors that could interact with cancers. However, many existing and clinical databases cannot be utilized due to the lack of available bioinformatics pipelines. Therefore, we implemented a methodology that investigated possible comorbidity interactions of COVID-19 with a number of cancers relating to breast, lung, colon, kidney, liver, prostate, bladder and thyroid by examining the gene expression profiling. This analysis has been used to combine gene expressions, gene ontology and molecular instances by manipulating gene set enrichment analysis (GSEA) and semantic similarity, respectively. Therefore, various significant genes, GO terms and pathways were determined as the proximities and identified a potential interacting biological process (BP) for each disease.

Materials and methods

Bioinformatics and integrative procedures [5] were used to investigate the relations among COVID-19 and various cancers that are described as follows:

Data collection

The experimental datasets were obtained from the Gene Expression Omnibus (GEO) database, National Centre for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov/geo/) and Array Express Database of European Bioinformatics Institute (EBI) (https://www.ebi.ac.uk/arrayexpress/). There were two query results found for COVID-19. Four main principles were used to identify appropriate microarray transcriptomics datasets, which are given as follows:

1. Redundancy: Several datasets are generated using similar conditions or explored with various methods. In these circumstances, no equivalent samples are not required more than one time.

2. Typology: Datasets required more accurate structural form such as sequential data.

3. Relevance: Datasets must be linked to specific pathology that gives certain importance about the biological relationships. Several samples does not contain its own pathology, hence they are imperfect for further analysis.

4. Species: Datasets must be gathered from clinical sources and not derived from non-human species.

Gene Set Enrichment Analysis

GSEA is a functional process where a group of genes, their enhanced expressions and the effects of case versus control tissues are identified using statistical approaches. Also, they are recognized by genes and protein set that associates with particular disease phenotypes based on similar biological functionalities, chromosomal location and regulation [6]. However, the transcriptomic and proteomic data are investigated in this condition. Further, DNA microarray or next-generation sequencing (NGS) data is explored by comparing genes from two cells or tissues and scrutinizing gene expressions depending on several states. These gene sets are interrelated with the phenotypic differences under the list of up- and down-regulated genes In this study, we gathered two COVID-19 and various cancer samples from GEO and EBI repository. The brief description of these datasets is given as follows (see Tables 1 and 2):

Table 1.

Selected COVID-19 datasets

Disease name	Dataset	Tissue/source	Control	Case
COVID-19	GSE147507 [7]	Human lung	3	3
	PBMC-COVID-19 [8]	Peripheral blood	3	3

Table 2.

Selected cancer datasets

Disease name	Dataset	Tissue/source	Control	Case
Breast cancer (BC)	GSE98528 [9]	Invasive lobular carcinoma	9	39
	GSE107300 [10]	Lung metastatic subline	6	6
	GSE110332 [11]	Breast cancer cell SUM159	3	3
	GSE124646 [12]	Breast biopsy	10	90
	GSE125989 [13]	Breast biopsy	16	16
Colon cancer (CC)	GSE78051 [14]	Colorectal cancer cells	3	3
	GSE92921 [15]	Colon tissue biopsy	33	26
	GSE94154 [16]	Colorectal adenocarcinoma cells	3	3
	GSE110425 [17]	Colon cancer cell	6	6
	GSE115716 [18]	pN1-LS174T cells	9	18
Kidney cancer (KC)	GSE105261 [19]	Kidney biopsy	9	35
	GSE117890 [20]	Kidney tissue	6	5
Liver cancer (LC)	GSE63067 [21]	Liver tissue	7	11
	GSE102079 [22]	Liver tissue	14	243
Bladder & prostate cancer (BPC)	GSE118123 [23]	Prostate cancer cell	3	3
	GSE122306 [24]	Bladder cancer cell	6	6
Thyroid cancer (TC)	GSE3678 [25]	Thyroid	7	7
	GSE65144 [26]	Thyroid	13	12
	GSE85457	Thyroid	3	4

Pathway

Molecular pathways are perturbed in diseased conditions and identification of them enriched by the DEGs provides critical signaling pathways and drug targets. We utilized KEGG database [27] to identify COVID-19 pathways overlapped with different cancers enriched by the DEGs.

Ontology

GO is a conceptual model where biological information can be explored as a compatible and widespread structure. It represents genes and their related attributes across all species. The main purpose of GO is to represent, maintain, develop and annotate gene and gene products in details. Three GO domains are considered such as BP, molecular function and cellular component. However, pathological processes, experimental conditions and temporal information are not captured properly in this process. Alternatively, disease ontology (DO) denotes an open-source model that represents expansive information about inherited, developmental and acquired human diseases [28]. In this study, DO terms were extracted for the corresponding diseases such as COVID-19 DO ID: 00080600, breast cancer DO ID: 1612, colon cancer DO ID: 219, obesity DO ID: 9970, liver cancer DO ID: 3571, kidney cancer DO ID: 263, thyroid gland cancer DO ID: 1781, urinary bladder cancer DO ID: 11054 and prostate cancer DO ID: 10283. These DO IDs were retrieved from https://disease-ontology.org/. But, the result of SARS-CoV-2 is not available, hence we used DO ID of SARS coronavirus to compare DO with others.

Semantic similarity

Semantic similarity is a function that measures the proximity between two terms annotating to the biological entities on a given ontology. Numerous methods are employed to organize common ancestor terms in view of the annotation statistics. In this work, the relations included more significant terms among genes, GO and DO than particular evaluations. The Wang method fits in this purpose because graph-based method constructs the topology and inherits by the selected ontology.

A directed acyclic graph is defined as , where GO term , the set of ancestor terms and edges (semantic relations) are related to . The semantic value is manipulated as

(1)

where and specifies the generic and child term individually. According to the relation, the semantic contribution is assigned as 0 and 1 between and and global semantic value for is calculated in Eq. 2.

(2)

If and are measured using two terms and , then the semantic similarity is given in Eq. 3.

(3)

Given are two term sets and , where and are denoted as the length of the first and second set, respectively. The best-match average (BMA) method [29] generates the semantic similarity between the two sets (see Eq. 4):

(4)

with , indices on , terms.

Designing of pipeline

Figure 1 shows the steps of the pipeline:

Figure 1.

Working pipeline.

Table 5.

Summary of results along with the pipeline steps for the selected pathologies. The features from left to right are denoted as selected disease, source, number of data sets, number of selected data sets and number of upregulated and downregulated DEGs

Disease	Origin/tissue	Dataset	Selected dataset	DEG up	DEG down
COVID-19	Lung and blood	2	2	1239	614
Breast cancer	Breast	260	5	5603	6443
Colon cancer	Colon	120	5	372	674
Kidney cancer	Kidney	80	2	7145	7613
Liver cancer	Liver	80	2	788	718
Thyroid cancer	Thyroid	240	3	15004	13187
Bladder & prostate	Bladder & prostate	80	2	11	8

1. In the data extraction process, the selected COVID-19 and cancer datasets were downloaded and explored matrix information. The normalization was performed to convert them into expression classes. Subsequently, DEGs were identified as a linear and Bayesian method by comparing the expression of healthy controls or treated COVID-19 patients.

2. These samples were manually gathered to conduct this work. Then, we reviewed, selected and classified GEO samples (GSM) very meticulously rather than the automatic selection process.

3. Differential expression can be used for identifying significant genes altered in a particular condition. To identify DEGs, a linear and Bayesian method was applied [27]. We considered three statistical criteria, namely P-value, adjusted P-value (False Discovery Rate) and absolute logFC values to screen statistically significant DEGs.

4. In the GO term test, the class called topGOdata was created, which picked GO terms and genes to implement filtering function. The mapping had been engaged for annotation where Fisher’s exact test was used to explore the relationship between GO terms and genes.

5. After the mapping of the semantic similarity, the performance among all the selected pathologies were compared by means of genes, GO terms, DO terms for discrimination of the intimacy among the designated datasets.

6. Cluster comparison was used to fetch significant pathologies and enrichment test was dependent on DEGs and KEGG pathways for COVID-19 and cancers.

7. Finally, the output of this process provided a statistical summary, genes-GO terms, GO graph topology, gene semantic similarity matrix (and dendrogram), GO semantic similarity matrix (and dendrogram), DO semantic similarity matrix (and dendrogram), KEGG enrichment graph and the list of the common pathways pathologies [30]. In addition, the list of DEGs constructed gene networks corresponding to the information related to pathways/pathologies.

Then, we represented this work using two R scripts that are available at https://github.com/shahriariit/COVID-Cancer-Comorbidities. To build this bioinformatics pipeline, we used several Bioconductor packages [31] such as: “GEOquery” [32] for downloading GEO data and transformation of expression set class; “LIMMA” [33] for microarray data analysis, linear models and identifying DEGs on microarray data; “genefilter” [34] for keeping basic tasks of filtering genes; “topGO” for verifying GO terms and topology of DAG; “GOSemSim” [35] for the semantic similarity assessment among the diseases; “DOSE” [36] for the semantic similarity assessment among DO terms; and “clusterProfiler” [37] for the enrichment analysis with KEGG pathways.

Results

Statistical analysis of transcriptomic data

To identify common dysregulated DEGs between COVID-19 and cancers, we comprehensively analyzed the available transcriptomics datasets. The statistical summary of the COVID-19 and their cancer comorbidities have been presented in Tables 3 and 4, respectively. The th and th column of Table 4 provides the number of DEGs that retains the statistical threshold of P-value . We extracted up- and down-regulated genes based P-value and absolute fold change () for GSEA analysis where denotes the direction of gene expression. In the th column, the number of significant DEGs is presented that gives the specific threshold and used them for GO mapping. In the th column, the number of annotated GO terms of DEGs are provided. Later, Fisher’s exact test was employed to extract statistically significant terms based on gene counting. The classical enrichment analysis was performed to evaluate the over-representation of these terms within the DEGs group. We summarize GO terms in the last column of Tables 3 and 4. For example, GO graph on GSE147507 represents the hierarchy and zoom on significant GO terms at Figure 2.

Table 3.

Statistical summary of COVID-19 datasets used in this study. Columns , , and represent the number of unfiltered genes, the number of significant DEG with threshold for P-value, adjusted P-value and logFC, respectively. Columns and show the number of raw GO terms and significant GO terms with Fisher test, respectively.

Dataset	Source tissue	Raw genes	P-value	Adjusted P-value	logFC	GO terms	Fisher test
GSE147507	Human lung	108	108	108	108	1117	446
PBMC-COVID-19	Peripheral blood
mononuclear cells		1745	1745	1745	1745	4386	1745

Table 4.

Statistical summaries of cancer comorbidities datasets. Columns , , and specify the number of unfiltered genes, the number of significant DEG with the threshold for P-value, adjusted P-value and logFC, respectively. Columns and show the number of raw GO terms and significant GO terms by Fisher test, respectively. Data set legend: BC-GSE98528, GSE107300, GSE110332, GSE124646 and GSE125989; CC-GSE78051, GSE92921, GSE94154, GSE110425, GSE115716; KC-GSE105261 and GSE117890; LC-GSE63067 and GSE102079; BPC-GSE118123 and GSE122306; and TC-GSE3678, GSE65144 and GSE85457

Dataset	Source tissue (case control)	Raw genes	P-value	Adjusted P-value	LogFC	GO terms/ raw GSEA	Fisher test
GSE118123	Prostate cancer cell	54675	4255	6	2	172	56
GSE122306	Bladder cancer cell	54675	3086	6	17	133	70
GSE98528	Invasive lobular carcinoma	46446	3522	0	50	127	63
GSE107300	Lung metastatic subline	47302	8500	3326	8816	100	55
GSE110332	Breast cancer cell SUM159	22277	3340	594	58	170	70
GSE124646	Breast biopsy	22283	5878	3183	1005	134	52
GSE125989	Breast biopsy	22277	1778	104	2097	122	31
GSE78051	Colon cancer cell	47323	3642	77	0	68	38
GSE92921	Colon tissue biopsy	54675	8359	1558	525	150	90
GSE94154	Colorectal adenocar-cinoma cells	54675	12141	4360	328	103	72
GSE110425	Colon cancer cell	47323	1743	0	1	133	60
GSE115716	pN1-LS174T cells	47323	6438	360	171	164	27
GSE105261	Kidney biopsy	48107	8090	2359	816	159	54
GSE117890	Kidney tissue	47309	5905	56	13925	212	142
GSE63067	Liver tissue	54676	4448	0	227	248	168
GSE102079	Liver tissue	54613	16539	8578	1263	179	74
GSE3678	Thyroid biopsy	54675	6290	1615	1215	227	121
GSE65144	Thyroid biopsy	54675	16368	10748	10911	151	74
GSE85457	Thyroid biopsy	54613	7777	297	16055	329	189

Figure 2.

Example of the GO graph with GSEA on GSE147507 data set. The rectangles represent the top five GO terms after the test. The red and orange colors indicate the most significant GO terms.

KEGG pathway

To clarify the significance of the DEGs from transcriptomic datasets, we have performed gene ontologies and pathway analysis. The pathway-based analysis represents how complex diseases associates with other underlying molecular mechanisms [27]. Moreover, the following framework is provided on the BP involved in each COVID-19 study.

• GSE-147507: reproduction, MAPK cascade, angiogenesis, blood vessel development, cell activation

• PBMC-COVID-19: multicellular organismal process, developmental process, anatomical structure development, multicellular organism development, system development

GO enrichment and construction of GO terms tree

We compared the DEGs identified from genome-wide transcriptomic datasets of COVID-19 and selected cancers and identified several common dysregulated genes (MARCO, VCAN, ACTB, LGALS1, HMOX1, TIMP1, OAS2, GAPDH, MSH3, FN1, NPC2, JUND, CHI3L1, GPNMB, SYTL2, CASP1, S100A8, MYO10, IGFBP3, APCDD1, COL6A3, FABP5, PRDX3, CLEC1B, DDIT4, CXCL10 and CXCL8) that are found common between COVID-19 and cancer (see Figure 3). To provide insights into the functional interactions of the identified genes, a protein–protein interaction network is created around the common DEGs using GeneMania web-utility considering co-expression, physical interaction, pathway, co-localization, generic interaction, predicted and shared protein domains.

Figure 3.

Network on common differential expressed genes between COVID-19 and its cancer comorbidities.

Similarly, the common genes of COVID-19 (lung and blood tissues) and individual cancers are shown in Tables 6 and 7, respectively, which are obtained from the comparison between COVID-19 and cancer comorbidities.

Table 6.

Common GO terms among COVID-19 (lung) and cancers

GSE ID	GO ID	GO term	GSE ID	GO ID	GO term
Common GO terms between COVID-19 and breast cancer			Common GO terms between COVID-19 and lung cancer
BC1_GSE95165	GO:0002376	Immune system process	LC1_GSE63067	GO:0001775	Cell activation
BC4_GSE110332	GO:0002376	Immune system process		GO:0001816	Cytokine production
BC5_GSE124646	GO:0002376	Immune system process		GO:0001932	Regulation of protein phosphorylation
BC6_GSE125989	GO:0002376	Immune system process		GO:0001934	Positive regulation of protein phosphorylation
BC7_GSE135427	GO:0002376	Immune system process		GO:0002252	Immune effector process
	GO:0002520	Immune system development		GO:0002263	Cell activation involved in immune response
BC8_GSE89333	GO:0001568	Blood vessel development		GO:0002274	Myeloid leukocyte activation
	GO:0001775	Cell activation		GO:0002275	Myeloid cell activation involved in immune response
	GO:0001932	Regulation of protein phosphorylation		GO:0002366	Leukocyte activation involved in immune response
	GO:0001934	Positive regulation of protein phosphorylation		GO:0002376	Immune system process
	GO:0001944	Vasculature development		GO:0002443	Leukocyte mediated immunity
	GO:0002376	Immune system process		GO:0002444	Myeloid leukocyte mediated immunity
Common GO terms between COVID-19 and colon cancer				GO:0002446	Neutrophil mediated immunity
CC3_GSE92921	GO:0001775	Cell activation		GO:0002682	Regulation of immune system process
	GO:0001816	Cytokine production		GO:0002684	Positive regulation of immune system process
	GO:0001817	Regulation of cytokine production	LC2_GSE102079	GO:0002252	Immune effector process
	GO:0002376	Immune system process		GO:0002376	Immune system process
	GO:0002682	Regulation of immune system process		GO:0002682	Regulation of immune system process
	GO:0002684	Positive regulation of immune system process	Common GO terms between COVID-19 and thyroid cancer
Common GO terms between COVID-19 and kidney cancer			TC2_GSE3678	GO:0001775	Cell activation
KD2_GSE105261	GO:0001775	Cell activation		GO:0002376	Immune system process
	GO:0002252	Immune effector process		GO:0006955	Immune response
	GO:0002253	Activation of immune response		GO:0007166	Cell surface receptor signaling pathway
	GO:0002376	Immune system process	TC2_GSE3678	GO:0000165	MAPK cascade
	GO:0002682	Regulation of immune system process		GO:0001568	Blood vessel development
	GO:0002684	Positive regulation of immune system process		GO:0001775	Cell activation
KD2_GSE105261	GO:0001775	Cell activation		GO:0001932	Regulation of protein phosphorylation
	GO:0002376	Immune system process		GO:0001944	Vasculature development
	GO:0006955	Immune response		GO:0002274	Myeloid leukocyte activation
	GO:0007166	Cell surface receptor signaling pathway		GO:0002376	Immune system process
	GO:0009605	Response to external stimulus	TC3_GSE65144	GO:0001775	Cell activation
	GO:0010033	Response to organic substance		GO:0002376	Immune system process
KD3_GSE117890	GO:0001775	Cell activation		GO:0002682	Regulation of immune system process
	GO:0002376	Immune system process	TC4_GSE85457	GO:0001568	Blood vessel development
	GO:0002682	Regulation of immune system process		GO:0001944	Vasculature development
				GO:0002376	Immune system process
				GO:0002682	Regulation of immune system process

Table 7.

Common GO term among COVID-19 (PBMC) and cancers

GSE ID	GO ID	GO term	GSE ID	GO ID	GO term
Common GO terms between COVID-19 and breast cancer			Common GO terms between COVID-19 and colon cancer
BC1_GSE95165	GO:0002376	Immune system process	CC2_GSE79462	GO:0007275	Multicellular organism development
	GO:0007275	Multicellular organism development		GO:0009653	Anatomical structure morphogenesis
	GO:0032501	Multicellular organismal process		GO:0030154	Cell differentiation
	GO:0032502	Developmental process		GO:0032501	Multicellular organismal process
BC2_GSE98528	GO:1901564	Organonitrogen compound metabolic process		GO:0032502	Developmental process
	GO:0042221	Response to chemical		GO:0042221	Response to chemical
	GO:0030154	Cell differentiation		GO:0048513	Animal organ development
	GO:0048869	Cellular developmental process	CC3_GSE92921	GO:0001775	Cell activation
	GO:0032502	Developmental process		GO:0002376	Immune system process
BC3_GSE107300	GO:0010033	Response to organic substance		GO:0006955	Immune response
	GO:0032501	Multicellular organismal process		GO:0007166	Cell surface receptor signaling pathway
	GO:0042221	Response to chemical		GO:0007275	Multicellular organism development
BC4_GSE110332	GO:0002376	Immune system process		GO:0009605	Response to external stimulus
	GO:0006955	Immune response		GO:0010033	Response to organic substance
	GO:0007166	Cell surface receptor signaling pathway	CC4_GSE94154	GO:0002376	Immune system process
	GO:0007275	Multicellular organism development		GO:0006955	Immune response
	GO:0009605	Response to external stimulus		GO:0007166	Cell surface receptor signaling pathway
	GO:0010033	Response to organic substance		GO:0009605	Response to external stimulus
BC5_GSE124646	GO:0002376	Immune system process		GO:0010033	Response to organic substance
	GO:0006955	Immune response		GO:0032501	Multicellular organismal process
	GO:0007166	Cell surface receptor signaling pathway	CC5_GSE110425	GO:0010033	Response to organic substance
	GO:0007275	Multicellular organism development		GO:0042221	Response to chemical
	GO:0009653	Anatomical structure morphogenesis	CC6_GSE115200	GO:0007275	Multicellular organism development
	GO:0010033	Response to organic substance		GO:0030154	Cell differentiation
	GO:0030154	Cell differentiation		GO:0032501	Multicellular organismal process
	GO:0032501	Multicellular organismal process		GO:0032502	Developmental process
BC6_GSE125989	GO:0002376	Immune system process		GO:0042221	Response to chemical
	GO:0006928	Movement of cell or subcellular component	CC7_GSE115716	GO:0050896	Response to stimulus
	GO:0007166	Cell surface receptor signaling pathway		GO:0051716	Cellular response to stimulus
	GO:0007275	Multicellular organism development	Common GO terms between COVID-19 and kidney cancer
	GO:0007399	Nervous system development	KD1_GSE51571	GO:0006928	Movement of cell or subcellular component
BC7_GSE135427	GO:0002376	Immune system process		GO:0007275	Multicellular organism development
	GO:0007275	Multicellular organism development		GO:0009653	Anatomical structure morphogenesis
	GO:0009653	Anatomical structure morphogenesis	KD3_GSE117890	GO:0001775	Cell activation
BC8_GSE89333	GO:0001775	Cell activation		GO:0002376	Immune system process
	GO:0002376	Immune system process		GO:0006955	Immune response
	GO:0006955	Immune response		GO:0007166	Cell surface receptor signaling pathway
	GO:0007166	Cell surface receptor signaling pathway		GO:0007275	Multicellular organism development
Common GO terms between COVID-19 and breast cancer			Common GO terms between COVID-19 and lung cancer
BP1_GSE118123	GO:0009605	Response to external stimulus	LC1_GSE63067	GO:0001775	Cell activation
	GO:0010033	Response to organic substance		GO:0002376	Immune system process
	GO:0042221	Response to chemical	LC2_GSE102079	GO:0002376	Immune system process
BP2_GSE122306	GO:0007275	Multicellular organism development		GO:0006955	Immune response
	GO:0009605	Response to external stimulus		GO:0007275	Multicellular organism development
	GO:0010033	Response to organic substance		GO:0009605	Response to external stimulus
			Common GO terms between COVID-19 and thyroid cancer
			TC1_GSE3467	GO:0032501	Multicellular organismal process
			TC3_GSE65144	GO:0001775	Cell activation
				GO:0002376	Immune system process
				GO:0006928	Movement of cell or subcellular component
				GO:0006955	Immune response
				GO:0007166	Cell surface receptor signaling pathway
				GO:0007275	Multicellular organism development
				GO:0010033	Response to organic substance
			TC4_GSE85457	GO:0002376	Immune system process
				GO:0006928	Movement of cell or subcellular component
				GO:0006955	Immune response

Semantic similarity analysis of the KEGG pathways

We performed the semantic similarity of the pathways enriched by the DEGs in order to prioritize and evaluate their proximity. Figure 4 shows the semantic similarity matrix for DEGs of the selected pathologies. The COVID-19 (PBMC) is highly connected to BC5_GSE124646, BC4_GSE110332 and TC3_GSE65144 when the values of semantic similarity are above 0.7. While COVID-19 (lung data) is highly associated with LC1_GSE63067 at the same semantic similarity values. When we consider semantic similarity value above 0.6 and less 0.7, then COVID-19 (lung) and COVID-19 (PBMC) are associated with several cancers like LC2_GSE102079, KD2_GSE105261, CC4_GSE94154, BC6_GSE125989, BC2_GSE98528, TC2_GSE3678 and CC3-_GSE92921 individually. At 0.5 semantic similarity score, COVID-19 (PBMC) is related to LC1_GSE63067. This matrix showed that TC1_GSE3468, BC3_GSE107300, BC2_GSE98528 and BP1_GSE1181123 provided low semantic similarity value with other cancers.

Figure 4.

Semantic similarity matrix for differential expressed genes. The two-letter suffix before the GSE codes referred to the following: BC, breast cancer; CC, colon cancer; PMVC1 and PMVC2, COVID-19; KD, kidney cancer; LC, liver cancer; TC, thyroid cancer; and BP, bladder prostate cancer. The number after the two letters indicates the logFC threshold.

Figure 5 represents the semantic similarity matrix of GO terms. Over the value of 0.7 and less than 0.8, all datasets are found well-clustered among themselves except COVID-19 (lung) and LC1_GSE63067. When the semantic similarity value was 0.8 and less than 0.9, LC2_GSE102079, TC1_GSE3467, TC3_GSE65144 and KD2_GSE105261 are also well-clustered with several cancer pathologies. When the semantic similarity value was 0.9 or above 0.9, TC1_GSE3468, TC2_GSE3678, CD1_GSE893333, BC4_GSE110332 and KD2-_GSE105261 are represented well-clustered.

Figure 5.

Semantic similarity matrix of GO terms. The number after each pair of entries represent the logFC threshold.

Figure 6 shows DO terms for SARS-CoV where COVID-19, breast cancer, kidney cancer, liver cancer and thyroid cancer are related with 0.09 threshold. Again, colon cancer contains 0.07 similarity value that is less connected than others. Instead, DO terms for SARS-CoV-2 are not available in the DO repository where it shows blank values in the generated graph. Hence, we used terms of SARS-CoV in this work.

Figure 6.

Semantic similarity matrix for DO terms (SARS-CoV)

However, Figures 7 and 8 show KEGG pathway association with selected datasets. This analysis is useful to understand how complex diseases may be related to each other through their underlying molecular mechanisms [27]. It represents the relationships between KEGG pathways of COVID-19 and associated cancer data sets. These pathways enriched by DEGs are shown in the dot plot where each row represents them associated with COVID-19 and various cancers. The domination of genes is determined by the dimension of the circles in the pathway and the range of the circles is computed the statistical validation for P-value = 0.05.

Figure 7.

KEGG pathway enrichment analysis for COVID-19 lung tissues.

Figure 8.

KEGG pathway enrichment analysis for COVID-19 blood tissues.

Common recurring pathways between COVID-19 (lung) and others pathologies are found including viral protein interaction with cytokine and cytokine receptor, Toll-like receptor signaling pathway, Influenza A, prion diseases, cytokine–cytokine receptor interaction, Rheumatoid arthritis, IL-17 signaling pathway, TNF signaling pathway and NOD-like receptor signaling pathway, among others.

Discussion

Bioinformatics is a very important and fast-growing field that can investigate the cause and interaction of various diseases in the medical sciences. The main purpose of this work is to explore the association between COVID-19 and its cancer comorbidities to understand the complexities of cancer patients if they are infected by the SARS-CoV-2. The entire research process relies on the different methods and techniques used for knowledge extraction in bioinformatics. Therefore, we examined the most recent COVID-19 and numerous cancers transcriptomic data in the publicly accessible repositories. In this integrated bioinformatics framework, numerous packages were implemented from the Bioconductor repository using R. GSEA is used to study COVID-19 in terms of the pathways and different ontologies such as GO and DO terms. We also began this test from the set of DEGs and defined GSEA taking into account the most relevant GO terms. In order to show the proximity between different diseases according to chosen ontologies, the usefulness of semantics similarity was again used. Furthermore, GSM documents were noted manually and samples were divided into control and case instead of automatic selection of GEO samples. Then, we created models using manually curated datasets instead of the automatic selection with GEO samples. In order to show the proximity between different diseases according to chosen ontologies, we used semantic similarity approach again. Then, all results containing genes, GO and DO terms were compared to evaluate semantic similarity. There is still no effective method to define the functional similarities based on gene annotation information from dissimilar data sources. Hence, GO terms are effective to address the consistent explanations about genes in different data sources. Instead, DO provides an open source ontology for the incorporation of biomedical data in human disease. It produces a consistent description of gene products with disease perspectives for supporting functional genomics. Several metrics like P-value and logFC thresholds are used in this work. For the P-value of 0.05 and absolute logFC of 1, the variances among sets of DEGs and GO terms are extracted. Consequently, we determined KEGG pathway graph that showed the connectivity of COVID-19 and other diseases. Our analysis identified a number of common dysregulated genes between COVID-19 and cancers. Among the identified common genes, MARCO and OAS2 were identified as dysregulated in breast cancer as consistent with previous report [38, 39]. Previous studies suggested OAS2 as prognostic markers of breast cancer [39]. Another gene, VCAN was identified as a new prognostic gene in gastric cancer [40]. The critical role of ACTB was also found in lung cancer [41]. Overexpression of LGALS1 gene was reported in oral cancer and has been detected as key players for various tumor including prostate, thyroid, bladder and ovarian cancer [42]. Higher expression of HMOX1 gene was revealed in cancer corroborating our findings [43]. TIMP1 was established as anti-apoptotic roles in colon cancer and suggested that it might be critical for cell proliferation, invasion and metastasis of colon cancer [44]. Again, the rest of the identified genes has represented key roles in the development and progression of cancer as consistent with previous findings. In order to shed light on biological pathways commonly altered in COVID-19 and cancer, we identified several pathogenetic processes and molecular pathways that may potentially clarify the potential mechanisms of COVID-19 in cancer patients. Our study highlighted immune system processes and cytokine-mediated inflammations as key BPs of COVID-19 and cancer. The chronic inflammation has been recognized as a causative factor for the progression of cancer [45]. Immune systems, cytokines overproduction and cytokine-mediated signaling provided key features in lung inflammation in response to COVID-19 infections [46], which are consistent with our findings. This study identified wnt signaling pathways, IL-17 signaling pathway, TNF signaling pathways as key signaling pathways associated with COVID-19 and cancers which is consistent with previous reports that identified these altered pathways in COVID-19 [46]. Specifically, the “cytokine storm” seen in COVID-19 patients is the result of severe immune response by the host that deteriorate the conditions of the patients [46]. In line with this evidence, we may suggest the dysregulated immune systems play a critical role in COVID-19 patients with cancers. Several previous studies employed whole genome transcriptomic data, identified gene signatures and elucidated immunopathological features and potential marker focused on COVID-19 [46–52], which were consistent with our findings; However, molecular associations between COVID-19 and different cancers have not been found yet. For the first time, we elucidated molecular cell pathways shared between COVID-19 and cancer individually.

It demonstrated the likelihood of reusing the data available from the analytical perspective. For further research, various works related to comorbidities and transcriptomics have been published. However, owing to legal or ethical concerns they are not open to the media at all times. In this study, we represented the datasets with more cell types and resources that investigated robust results than single cells and resources. Several challenges were considered while developing this pipeline. Firstly, it was not only concerned about control versus patients but also scrutinized genetic variants to show the risk of this disease and its variants. Secondly, the standard of data is not similar in all cases. For instance, it took a lot of effort to prepare GEO series data. Therefore the microarray data quality (e.g., the arrayQualityMetrics package) was retained and ideal for semi-automated analysis. However, this approach provides an automated way of gathering, comparing and evaluating microarray data. In this study, we have implemented a comprehensive bioinformatics pipeline where several common pathogenetic processes are detected and shared by COVID-19 and cancer that may aid the clinicians and bench scientists to further dismantle the complex interconnections of the patients. The pipeline can also be used to investigate COVID-19 and other comorbidities, which is freely accessible for clinical researchers to use.

Conclusion

We have developed an R pipeline that incorporates bioinformatics methods to identify the infectome, diseasome and comorbidities relationship among the infections and diseases. In this study, a large set of transcriptomic datasets of COVID-19 and different cancers have been utilized and identified molecular associations between them using our developed pipeline. Our analysis showed common dysregulated genes shared between COVID-19 and cancers. We detected immune systems processes as major dysregulated pathways in COVID-19 and common cancers. Such study is also helpful in evidence-based guidelines on COVID-19 in patients with cancer as our suggested pipeline combines an integrated structure for discovering COVID-19 molecular pathways and various pathologies. Our pipeline can also be used for infectome, diseasome and comorbidities analysis of other diseases by using a large set of transcriptomic data. We are unable to test this technique with further records because of the lack of COVID-19 data, which will be available for the research on COVID-19 by the scientist. We now suggest to incorporate more genome-wide transcriptomic data once it will be available to get more comprehensive understanding of the COVID-19 in cancer comorbid patients. Our pipeline can be an enormous opportunity for clinicians and scientists to provide new insights into COVID-19 pathways in cancer patients despite constraints on the availability of more transcriptomic data.

Key Points

• This work developed a bioinformatics pipeline and has been applied to detect infectome, diseasome and comorbidities between COVID-19 and cancer diseases.

• Bioinformatics analysis of COVID-19 and its malignant comorbidities are required to evaluate their roles for clinical and further implications of COVID-19.

• Several approaches such as gene set enrichment analysis and semantic similarity are used to investigate COVID-19 and its malignant comorbidities in this work.

• Numerous transcriptomic datasets are explored common genes, gene ontology, DO and pathways.

Md. Shahriare Satu received his B.Sc. and M.Sc. degrees in information technology from Janagirnagar University in 2015 and 2017, respectively. He is currently working as a lecturer at the Department of Management Information Systems, Noakhali Science and Technology University, Noakhali, Bangladesh. From March 2016 to 2 December 2018, he worked as a lecturer at the Department of Computer Science and Engineering, Gono Bishwabidyalay, Bangladesh. His research interest includes data mining, health informatics and big data analytics.

Md. Imran Khan received his B.Sc. in computer science and engineering from Gono Bishwabidyalay in 2019. His research interest includes machine learning and bioinformatics.

Md. Rezanur Rahman is a lecturer (on study leave) Khwaja Yunus Ali University, Bangladesh. He is a biotechnologist and bioinformatician by training. He received MSc and a BSc in biotechnology and genetic engineering with outstanding academic results holding first position in the faculty in both degrees from Islamic University, Bangladesh. For the highest CGPA in the faculty, he received the most prestigious award “Prime Minister Gold Medal 2017” from Prime Minister Sheikh Hasina. He has published 26 research articles in internationally reputed journals. His research focuses on to understand the molecular mechanism of complex diseases utilizing transcriptomics, systems biology and bioinformatics.

Koushik Chandra Howlader was born in Jhalakathi, Bangladesh, in 1991. He received his B.Sc. and M.Sc. degree in information technology from Jahangirnagar University, Bangladesh. Then he took his internship program from Infosys Ltd., India. After coming back to India, he joined as a junior Java Programmer in the IBCS-Primax Software Ltd., Dhaka. Currently, he is working as an assistant professor at the Department of Computer Science and Telecommunication Engineering at Noakhali Science and Technology University, Bangladesh. From 27 December 2015 to 26 December 2017, he worked as lecturer in the same department Noakhali Science and Technology University. His research interest includes data mining, bioinformatics, machine learning, etc.

Shatabdi Roy received her B.Sc. in computer science and telecommunication engineering from Noakhali Science and Technology University in 2019, respectively. Her research interest includes bioinformatics and bio-medical engineering.

Shuvo Saha Roy received his B.Sc. in computer science and telecommunication engineering from Noakhali Science and Technology University in 2019, respectively. His research interest includes bioinformatics.

Julian M. W. Quinn received the PhD from the University of Oxford, UK, in 1992, then moved to Australia for postdoctoral training in bone, joint and cancer biology at the St Vincent’s Institute of Medical Research as a senior research fellow since 2014. His interests are in applications of biostatistics and bioinformatics, and now he works as a surgical research officer at the Surgical Education and Research Training Institute at Royal North Shore Hospital, Sydney Australia.

Mohammad Ali Moni is a research fellow and conjoint lecturer at the University of New South Wales, Australia. He received his PhD degree in clinical bioinformatics and machine learning from the University of Cambridge. His research interest encompasses artificial intelligence, machine learning, data science, health informatics and clinical bioinformatics.