Computational Analysis of Protein Interactions between HIV-1 and

Homo sapiens
Abstract:
In order to understand the mechanism of infection and develop better treatment and prevention of
infectious diseases, host-pathogen interactions are important. The protein interaction map will guide
research on key PPIs that may lead to human cells being adhered to, colonized, and even invaded by
pathogens. Host-pathogen PPI prediction, however, has its challenges.

Introduction:
Background

In the process of transmission, viruses like HIV-1 will enter a cell, migrate to locations where they can
hijack host machinery to transcribe their genes and express their proteins, combine and then exit the cell
again while evading the host immune system. Therefore, promising infection requires the ability of the
pathogen to manipulate the metabolic pathways and mechanisms of the infected organism. Interactions
between Aids-encoded and sapient proteins include one way that HIV-1 can link to these survival
mechanisms in cellular pathways.

Results

Host-pathogen interactions are important to understand the function of infection and develop better
treatment and prevention for infectious diseases. The protein interaction map should direct work on main
PPIs that can contribute to the compliance, colonization and even invasion of pathogens by human cells.
However, the prediction of host-pathogen PPI has its challenges. Approaches for assessing and analyzing
host-pathogen PPIs can be categorized into assessment based on gold standard PPIs functional
information analysis in terms of Gene Ontology, pathways, gene expression data and RNA interference
data. Localization information analysis in terms of protein sub-cellular localization and co-localization of
host and pathogen proteins related experimental data analyses and biological case studies and
explanations, some of these assessment approaches can also be used as filtering strategies for pruning
host-pathogen PPI prediction results.

Related Work:
(Zhao, Chen, Zhang, Yi, & Zhao, 2020) summarized multiple pathways for interaction with cancer.
Although several viruses have been proved to induce host specific microRNAs (miRNAs, miRs), the
expression of functional miRNAs induced by Human Immunodeficiency Virus 1 (HIV-1) infection is still
unknown. In this study, they attempted to establish a macro control regulation system and simulate the
influence of HIV-1 inducing miRNAs during the development of cancer. The miRbase, FunRich
software, miRtarbase, STRING, TargetScanhuman, Cytoscape plugin ClueGO/Cluepedia/STRING,
DAVID Bioinformatics Resources and GEO database were comprehensively employed in this
bioinformatics study. They concluded that the miRNAs in the serum of AIDS patients and its target genes
have different expression levels in serum, an array of which are associated with cancer and metabolism
signaling pathways. Moreover, the emerging role of miRNAs in HIV-1 infection is also involved in
human cancer, using TCGA data integrative analysis.Therefore, they infer that serum miRNAs in HIV-1
infection may play important roles in HIV-induced cancer and could be used as a potential biomarker for
HIV-cancers detection.
In this article, (Qi, Tastan, Carbonell, Klein-Seetharaman, & Weston, 2011) designed a semi-supervised
multi-task learning framework to integrate a larger set of potentially interacting protein pairs retrieved
from literature (weak labels) and a smaller set of interactions annotated by experts. The proposed SML
combine a semi-supervised auxiliary task with a supervised PPI classifier. A multi-layer perceptron
network is trained for PPI classification on labeled examples. Simultaneously, they multi-task this
network with an auxiliary task which aims to use weak positive labels to improve the supervised
classification. Three auxiliary strategies are evaluated on the task of predicting interactions between HIV-
1 and human proteins. Through CV, this method was shown to improve upon the best previous method
for this task indicating the benefits of multitasking with auxiliary information. In addition to improved
performance on inferring human HIV-1 PPIs, the proposed SML structure provides a flexible framework
for general computational PPI prediction tasks. The method has significant potential for intra-species PPI
predictions such as in human.

(Cui, Fang, & Han, 2012) developed a new method for representing a protein sequence of variable length
in a frequency vector of fixed length, which encodes the relative frequency of three consecutive amino
acids of a sequence. They built a support vector machine (SVM) model to predict human proteins that
interact with virus proteins. In two types of viruses, human papillomaviruses (HPV) and hepatitis C virus
(HCV), SVM model achieved an average accuracy above 80%, which is higher than that of another SVM
model with a different representation scheme. Using the SVM model and Gene Ontology (GO)
annotations of proteins, they predicted new interactions between virus proteins and human proteins.

(Doolittle & Gomez, 2010) developed and applied a computational approach to predict interactions
between HIV and human proteins based on structural similarity of 9 HIV-1 proteins to human proteins
having known interactions. Using functional data from RNAi studies as a filter, they generated over 2000
interaction predictions between HIV proteins and 406 unique human proteins. Additional filtering based
on Gene Ontology cellular component annotation reduced the number of predictions to 502 interactions
involving 137 human proteins. They find numerous known interactions as well as novel interactions
showing significant functional relevance based on supporting Gene Ontology and literature evidence.

The author (Bandyopadhyay, Ray, Mukhopadhyay, & Maulik, 2014) have reviewed the computational
approaches that are devoted to the analysis and prediction of HIV-1-human PPIs. They have broadly
categorized these studies into two fields: computational analysis of HIV-1-human PPI network and
prediction of novel PPIs. They have also presented a comparative assessment of these studies and
proposed methodologies for discussing the implication of their results. They have also reviewed different
computational techniques for predicting HIV-1-human PPIs and provided a comparative study of their
applicability. Their effort will provide helpful insights to the HIV research community.

In this study, authors (Dickerson, Pinney, & Robertson, 2010) described HIV's propensity to interact with
highly connected and central proteins is a consequence of interactions with particular cellular functions,
rather than being a direct effect of network topological properties. The lack of a propensity for
interactions with phenotypically essential proteins suggests a selective pressure to minimize virulence in
retroviral evolution. Thus, the specificity of HIV-host interactions is complex, and only superficially
explained by network properties.

In this paper, (Lestari, Aprilia, & Bustamam, 2018) proposed technique for detecting protein-protein
interaction (PPI) networks using Global Encoding (GE) as feature extraction and combined with Support
Vector Machine (SVM) as a classifier. The experimental results show that GE approach is convenient to
calculate and effective in feature extraction problem for amino acid sequences. The application of Support
Vector Machine (SVM) predictor ensures reliable recognition with minimum error. Experimental results
showed that the method they used performed significantly well in distinguishing interacting and non-
interacting protein pairs with HIV-1. Their method performed better performance in the larger linear
kernel in terms of classification accuracy for PPI (the best accuracy of 85% with linear kernel L = 8).
Moreover, this PPI study can extend in order to find the right drug for people with HIV-1.

(Yang, Fu, Lian, Dong, & Zhang, 2019) have found that viral targets are enriched within human protein
complexes. They observed in the context of VTCs that viral targets tended to have a high within-complex
degree and to be scaffold and housekeeping proteins. Complexes that are essential for viral propagation
were simultaneously targeted by multiple viruses. They characterized the periodic expression patterns of
VTCs and provided the corresponding candidates that may be involved in the manipulation of the host
cell cycle. As a potential application of the current analysis, they proposed a VTC-based antiviral drug
target discovery strategy.

In this study, (Barman, Saha, & Das, 2014) a systematic attempt has been made to predict viral-host PPIs
by integrating different features, including domain-domain association, network topology and sequence
information using viral-host PPIs from VirusMINT. The three well-known supervised machine learning
methods, such as SVM, Naı ¨ve Bayes and Random Forest, were employed to evaluate the performance
measure based on five-fold cross validation techniques.Out of 44 descriptors, best features were found to
be domain-domain association and methionine, serine and valine amino acid composition of viral
proteins. In this study, SVM-based method achieved better sensitivity of 67% over Naı ¨ve Bayes
(37.49%) and Random Forest (55.66%). However the specificity of Naı ¨ve Bayes was the highest
(99.52%) as compared with SVM (74%) and Random Forest (89.08%). Overall, the SVM and Random
Forest achieved accuracy of 71% and 72.41%, respectively. The proposed SVM-based method was
evaluated on blind dataset and attained a sensitivity of 64%, specificity of 83%, and accuracy of 74.

This paper (Tastan, Qi, Carbonell, & Klein-Seetharaman, 2009) extended methods for predicting
interactions between hosts and pathogens, here human and HIV-1. Features derived from multiple
genomic and functional data sources and exploiting our knowledge of the human protein interactome
were integrated in a supervised learning framework. They inspected their predictions with independent
biological datasets to identify the most promising interacting pairs. By providing these new testable
hypotheses their predictions will accelerate experimental efforts to define a reliable network of HIV-l,
human protein interactions.

Experimental Results:
SVM:
precision recall f1-score support

CYT 0.34 0.84 0.48 94

EXC 0.00 0.00 0.00 8
ME1 0.00 0.00 0.00 5
ME2 0.00 0.00 0.00 12
ME3 0.00 0.00 0.00 31
MIT 0.35 0.16 0.22 45
NUC 0.40 0.18 0.25 95
POX 0.00 0.00 0.00 2
VAC 0.00 0.00 0.00 5

accuracy 0.35 297

macro avg 0.12 0.13 0.10 297
weighted avg 0.29 0.35 0.26 297

Conclusion:
We presented research on PHI networks ' bioinformatics review. Integrative study of high-throughput
omics studies utilizing computational methods will not only clarify the pathways of disease, but will also
help to identify new therapeutic targets and medicines by specifically defining important pathogen genes,
enzymes, and metabolites. Despite the recent attempts discussed above, it is still in its infancy to use
process biology methods to analyze PHI structures, largely due to data scarcity. Continuing field studies
can contribute to more robust PHI networks in the coming decade to develop PHI-based infectious
disease solutions.

Future Direction:
A broad overview of the analysis of PINs of infectious diseases caused by viruses, bacteria and protozoan
parasites has been outlined in this review. It involves the various methodologies that researchers can
adopt while attempting to analyze such networks.

References:
Bandyopadhyay, S., Ray, S., Mukhopadhyay, A., & Maulik, U. (2014). A review of in silico approaches
for analysis and prediction of HIV-1-human protein-protein interactions. Briefings in
Bioinformatics, 16(5), 830–851. https://doi.org/10.1093/bib/bbu041
Barman, R. K., Saha, S., & Das, S. (2014). Prediction of interactions between viral and host proteins
using supervised machine learning methods. PLoS ONE, 9(11).
https://doi.org/10.1371/journal.pone.0112034
Cui, G., Fang, C., & Han, K. (2012). Prediction of protein-protein interactions between viruses and
human by an SVM model. BMC Bioinformatics, 13(SUPPL.7). https://doi.org/10.1186/1471-2105-
13-S7-S5
Dickerson, J. E., Pinney, J. W., & Robertson, D. L. (2010). The biological context of HIV-1 host
interactions reveals subtle insights into a system hijack. BMC Systems Biology, 4.
https://doi.org/10.1186/1752-0509-4-80
Doolittle, J. M., & Gomez, S. M. (2010). Structural similarity-based predictions of protein interactions
between HIV-1 and Homo sapiens. Virology Journal, 7, 1–15. https://doi.org/10.1186/1743-422X-
7-82
Lestari, D., Aprilia, S., & Bustamam, A. (2018). Performance analysis of support vector machine
combined with global encoding on detection of protein-protein interaction network of HIV virus.
AIP Conference Proceedings, 2023. https://doi.org/10.1063/1.5064225
Qi, Y., Tastan, O., Carbonell, J. G., Klein-Seetharaman, J., & Weston, J. (2011). Semi-supervised multi-
task learning for predicting interactions between HIV-1 and human proteins. Bioinformatics, 27(13),
i645–i652. https://doi.org/10.1093/bioinformatics/btq394
Tastan, O., Qi, Y., Carbonell, J. G., & Klein-Seetharaman, J. (2009). Prediction of interactions between
HIV-1 and human proteins by information integration. Pacific Symposium on Biocomputing 2009,
PSB 2009, 516–527. https://doi.org/10.1142/9789812836939_0049
Yang, S., Fu, C., Lian, X., Dong, X., & Zhang, Z. (2019). Understanding Human-Virus Protein-Protein
 Interactions Using a Human Protein Complex-Based Analysis Framework. MSystems, 4(2), 1–16.
https://doi.org/10.1128/msystems.00303-18
Zhao, T., Chen, W., Zhang, X., Yi, H., & Zhao, F. (2020). HIV-induced cancer–all paths leading to
Rome. Microbial Pathogenesis, 139, 103804. https://doi.org/10.1016/j.micpath.2019.103804