مشاريع طلاب ربيع 2024 _ S24
DNA Methylation Patterns of Behavioural Disorders: A Cross-trait Analysis
DNA methylation is a key mechanism in epigenetics, influenced by environmental factors and disorders that cause changes in methylation patterns across DNA strands. These changes can affect phenotype without altering nucleotide sequences, often silencing genes through hypermethylation at promoter regions. Persistent methylation alterations may lead to mutations, prompting researchers to study DNA methylation biomarkers associated with diseases, with cancers being the most extensively investigated. Other disorders receive much less attention, and so do environmental factors. The motive of this study originates from the principle that prevention is better than treatment, particularly when the mechanisms of certain disorders remain unclear, with no established risk factors. Psychiatric conditions and behavioural disorders exemplify this issue, as most lack definitive treatments. Moreover, the diversity and overlap of symptoms complicate diagnosis further. Available medications for these conditions often come with side effects, especially when used long-term. This has inspired us to explore potential environmental contributions to such disorders, with the goal of improving individuals' quality of life without necessarily relying on medication, or at least minimizing its use. In addition, identifying potential environmental factors or epigenetic biomarkers can raise awareness about disease prognosis and ultimately help reduce incidence rates. Micro array data for 9 different phenotypes (Total 145 Subjects) were analysed for significant DMRs, with the results being gene-enriched prior to cross-comparison. From a technical standpoint, a comprehensive literature review on data preprocessing methods was conducted. Consequently, the tools used for analysis were selected based on recent advancements and literature recommendations. SeSAMe from R Bioconductor was employed for its modern p-value calculation method and comprehensive QC masking. This was followed by additional customized QC steps to minimize imputation of masked values and ensure high data quality. DMRs were detected using the Limma package, with FDR correction applied. While the alteration in methylation levels detected in this study was generally limited to single-probe differentiation, several DMPs were shared at the gene level. Both CD81 and MAD1L1 exhibit hypomethylation associated with DLD and one or two Alzheimer's groups. Interestingly, CD81 has been reported as upregulated in Alzheimer's candidates in a study using prefrontal cortex tissue samples. Therefore, our research, along with other studies, provides further evidence supporting the potential of peripheral blood biomarkers in reflecting neurological symptomatology. MAD1L1, on the other hand, has been frequently discussed in existing literature regarding the methylation of the same gene in psychiatric and environmental contexts. These findings should encourage further investigation of MAD1L1 to explore its potential role in neuropsychiatric symptoms. The study also emphasizes the need for standardized methods tailored to specific cell types or phenotypes. Such standardization would improve result consistency and enhance the reliability of DNA methylation analysis, particularly for diseases that lacks global methylation changes.
إعداد: الطالب سعيد عمر سعدو
إشراف: الدكتور ينال أحمد القدسي
DNA Methylation Patterns of Behavioural Disorders: A Cross-trait Analysis
Gene Expression Responses to Viral Infection Using RNA-Sequencing: Differential Expression Analysis in SARS- CoV-2-Infected Organoids
COVID-19, caused by the SARS-CoV-2 virus, is a complex, multi-organ disease with effects extending beyond the respiratory system to other vital organs such as the heart. This thesis investigates the transcriptomic responses in different tissues to uncover the molecular mechanisms driving COVID-19’s systemic impact. RNA sequencing (RNA-seq) was used to systematically profile transcriptional changes of hPSC-derived cells/organoids caused by SARS-CoV-2 infection. This study investigates the multiorgan impact of COVID-19. Using RNA sequencing, gene expression profiles were generated from cardiac and airway tissue samples obtained from infected and mock-treated controls. RNA-Seq analysis involved key steps such as Quality Control, Alignment to a reference genome, Gene quantification and Differential expression analysis to identify differentially expressed genes (DEGs). Broad differential expression datasets were generated to explore transcriptional responses, providing a comprehensive overview of gene expression changes. Visualizations such as volcano plots and heatmaps highlighted key patterns, guiding further refinement. To ensure highconfidence results, stricter thresholds were applied, narrowing the datasets to focus on biologically significant genes. The results revealed distinct patterns of upregulated and downregulated genes in both tissues, with a subset of overlapping DEGs suggesting shared pathways contributing to systemic inflammation. Tissue-specific analysis highlighted the enrichment of pathways related to immune response, cytokine signaling, and cell death in airway samples, while cardiac tissue exhibited significant enrichment in pathways associated with oxidative stress, fibrosis, and mitochondrial dysfunction. Venn diagram analysis further emphasized the tissue-specific and overlapping gene expression changes, providing insight into the multi-faceted nature of COVID-19 pathology. This study highlights the utility of RNA-Seq in elucidating molecular mechanisms underlying COVID-19 as a multi-organ disease, offering valuable insights into potential therapeutic targets for mitigating its systemic effects.
إعداد: الطالب عبد الحميد الخطيب
إشراف: الدكتور مجد الجمالي
تصميم حاسوبي للقاح متعدد الحواتم ضد فيروس انفلونزا الطيور باستخدام أدوات المعلوماتية المناعية
IN-SILICO DESIGN OF MULTI-EPITOPE VACCINE FOR H5N1 VIRUS USING IMMUNOINFORMATIC TOOLS
The highly pathogenic avian influenza virus (H5N1) remains a persistent threat to global public health and poultry industries due to its zoonotic potential, rapid mutation rates, and antigenic variability. Traditional vaccine development strategies are often limited by time constraints, high costs, and inadequate cross-strain protection.
In this study, an immunoinformatics-driven approach was employed to design a novel multi-epitope vaccine against H5N1. Viral protein sequences for Hemagglutinin (HA) and Neuraminidase (NA) were retrieved from the UniProt database and subjected to comprehensive epitope prediction and screening for T-cell (MHC-I and MHC-II) and B-cell (linear and discontinuous) epitopes. The selected epitopes were filtered based on their antigenicity, non-allergenicity, non-toxicity, and cytokineinducing potential. A multi-epitope vaccine construct was designed by assembling these epitopes with appropriate adjuvants (MDA5, H9E) and linkers (AAY, GPGPG, KK) to ensure immunogenicity and structural stability. Physicochemical analysis predicted a stable and hydrophilic vaccine construct, suitable for bacterial expression systems. Tertiary structure modeling using RoseTTAFold and subsequent validation (Ramachandran plot, ERRAT, ProSA) confirmed the reliability and stability of the protein model. Molecular docking studies revealed strong binding interactions with Toll-like receptors TLR7 and TLR8, indicating potential immune activation pathways. Immune simulation analysis predicted robust humoral and cellular immune responses, with increased IgG, IgM, T-helper cell activity, and IFN-γ cytokine production. Codon optimization and in-silico cloning into the pET-26b(+) vector confirmed efficient expression compatibility with Escherichia coli systems. The results suggest that the designed multi-epitope vaccine construct holds significant potential to elicit broad-spectrum immunity against H5N1 and overcome limitations associated with traditional vaccines. Future studies involving in-vitro and in-vivo validation are essential to confirm the vaccine's immunogenicity, safety, and efficacy.
إعداد: الطالب هشام الخطيب
إشراف: الدكتور ينال القدسي
IN-SILICO DESIGN OF MULTI-EPITOPE VACCINE FOR H5N1 VIRUS USING IMMUNOINFORMATIC TOOLS
Study of the docking of follicle stimulating hormone (FSH) to its normal receptor compared to the mutant type of the receptor
دراسة إرساء الهرمون المنبه للجريب FSH مع مستقبله الطبيعي مقارنة بالنمط بالطافر للمستقبل
FSHR is a critical G protein-coupled receptor essential for reproductive health, mediating hormonal signaling for oocyte and sperm development. Polymorphisms like Ser680Asn and Ala307Thr have been linked to male infertility by disrupting receptor-ligand interactions and signaling pathways. Bioinformatics and docking studies provide valuable insights into these mutations, paving the way for therapeutic strategies to address reproductive disorders and related diseases. This study explores the role of Follicle Stimulating Hormone (FSH) and its receptor (FSHR), focusing on the Ser680Asn and Ala307Thr polymorphisms linked to male infertility. FSHR variants were modeled using SWISS-MODEL, and transmembrane helices were predicted via TMHMM to identify functional domains. Docking simulations with HDOCK and interaction profiling using PLIP provided insights into the structural and binding dynamics of the FSH-FSHR complex. These findings highlight critical molecular interactions and the potential impacts of genetic polymorphisms on receptor functionality, offering a foundation for targeted therapeutic strategies. Our study demonstrated that the Ser680Asn polymorphism enhances ligand-receptor binding affinity compared to the wild-type receptor, potentially due to its role in stabilizing the extracellular binding pocket. However, the observed issue in this polymorphism may lie in the intracellular signal transduction process rather than the binding quality. Conversely, the Ala307Thr polymorphism showed reduced binding affinity, attributed to structural disruptions in the transmembrane domain that may impair receptor stability. The consistently high confidence scores support the reliability of the predicted interactions, with Ser680Asn exhibiting the strongest interaction profile. Structural flexibility, as assessed by variability in ligand positioning, appeared comparable across all receptor variants. These findings suggest that Ser680Asn may alter intracellular signaling efficiency, while Ala307Thr could contribute to reduced receptor functionality, with potential implications for reproductive health.
إعداد: الطالب فداء أحمد الطالب
إشراف: الدكتورة آية طوير
In Silico Design of a Novel Multi-Epitope Vaccine for Personalized Immunotherapy in Nasopharyngeal carcinoma (NPC)
Aim of the Study: This study aims to design and evaluate a novel multi-epitope vaccine targeting key antigens associated with NPC, specifically EBV latent proteins (EBNA1, LMP1, LMP2A) and the tumor-associated protein survivin. Utilizing advanced in silico approaches, the research seeks to identify highly immunogenic epitopes capable of eliciting strong cellular and humoral immune responses. The vaccine construct integrates these epitopes into a rationally designed framework, incorporating adjuvants, linkers, and structural elements to optimize immunogenicity, stability, and population coverage. Ultimately, the research aims to develop a targeted immunotherapeutic strategy that addresses the challenges of late-stage NPC, enhances treatment precision, and improves patient outcomes by offering a robust, safe, and effective vaccination platform.
Results: The study successfully designed a multi-epitope vaccine targeting EBV-associated antigens (EBNA1, LMP1, LMP2A) and survivin, critical for NPC progression. Computational analysis identified epitopes with strong immunogenicity, non-toxicity, non-allergenicity, and broad conservation across populations. The finalized vaccine construct demonstrated excellent solubility (SolPro score: 0.842; Protein-sol score: 0.629), good antigenicity (VaxiJen score: 0.550, ANTIGENpro score: 0.641), global coverage rate of 99.96% and a stable physicochemical profile, with a molecular weight of 56 kDa, theoretical isoelectric point (pI) of 9.83, an instability index of 31.86 classifying it as stable and a grand average of hydrophilicity (GRAVY) score of −0.376 suggesting hydrophilicity and enhanced solubility in aqueous environments. Codon optimization yielded a Codon Adaptation Index (CAI) of 1.0 which is ideal and a GC content of 49.09%, ensuring efficient expression in Escherichia coli (E.coli). ..........
إعداد: الطالبة سوسن علوان
إشراف: الدكتور مجد الجمالي
Evaluating Machine Learning Algorithms for Diabetes Prediction
The aim of this research is to identify high-performing machine learning algorithms capable of accurately predicting diabetes outcomes based on a wide range of clinical, demographic, and historical variables. These algorithms are designed to support early detection of diabetes and aid healthcare professionals in designing personalized treatment protocols, ultimately improving patient care and reducing diabetes-related complications Materials and Methods: This study utilized a dataset comprising information on 5,437 patients with 14 independent features representing demographics, clinical parameters, and medical history. The features included age, gender, 13 pulse rate, blood pressure (systolic and diastolic), glucose level, BMI, and family history of diabetes and related conditions such as hypertension and cardiovascular disease. Each record was labeled with a binary outcome indicating the presence or absence of diabetes. Data preprocessing steps included handling missing values, encoding categorical variables, and normalizing numeric attributes..........
إعداد: الطالبة هنادي المطيط
إشراف: الدكتورة آية طوير
Evaluating Machine Learning Algorithms for Diabetes Prediction
Automated Blood Cell Count using a Neural Network based on a Novel U-Net Architecture and Neubauer Haemocytometer
Large and expensive analytical machines pose an insurmountable hurdle to providing healthcare in remote suburban and rural areas. Such machines have proven to be cost prohibitive, immobile and locked down to specific vendors or countries. With advancement in machine learning algorithms and improved compute units a new chimeric option between manual slow and inaccurate analysis and large cost prohibitive immobile analysis machines. This option can prove to be a solution with minimal disadvantages. The objective of this paper is to develop a machine learning algorithm/neural network capable of carrying out blood cells count using a single image from a blood sample on a Neubauer counting chamber/slide. Developing such algorithms and models could pave the path for a universal and readily available on-the-fly blood tests in rural and suburban areas. Methods: a methodolgy based on heavy mathematical preprocessing of the images followed by UNet have been developed as a special architecture for the segmentation of medical images offering less computational load shifting the bulk of the processing onto mathematical models reducing the overall computational cost of the process. Results: the trained model has shown remarkable accuracy when it comes to segmentation (97.59%). However instance segmentations performance of stacked cells is yet insufficient for medical use.
إعداد: الطالب حيدر عمر عبد الدائم
إشراف: الدكتور حيان حسن
Automated Blood Cell Count using a Neural Network based on a Novel U-Net Architecture and Neubauer Haemocytometer
استقصاء تأثير الفيروسات التنفسية على التهاب المفاصل الروماتويدي باستخدام مقاربات حاسوبية
Investigating the Influence of Respiratory Viruses on Rheumatoid Arthritis (RA) Using In Silico Approaches
Rheumatoid arthritis (RA) is a chronic autoimmune disorder characterized by joint inflammation and systemic immune dysregulation. Viral infections, particularly respiratory viruses, have been implicated in triggering RA exacerbations, potentially through shared genetic and molecular pathways. In this study, we investigated the common molecular mechanisms between RA and a set of respiratory viruses, including Influenza, SARS-CoV-2, RSV, and others, using in silico methods. RNA-sequencing data from RA patients and viral infections were analyzed to identify differentially expressed genes (DEGs). Gene Ontology (GO) and KEGG pathway enrichment analyses revealed significant overlaps in biological processes, cellular components, and molecular functions, particularly in pathways related to immune response, cytokine signaling, and inflammation. Key pathways, such as the Neutrophil extracellular trap formation and Th1 and Th2 cell differentiation, were enriched in shared DEGs. Hub genes, including TLR2, IL6, S100A12, and IFI6, were identified, highlighting their central roles in both RA and viral immune responses. These findings provide insights into the molecular mechanisms underlying virus-induced RA flares and suggest potential therapeutic targets for mitigating disease exacerbations in RA patients during viral infections.
إعداد: الطالب ميلاد اكرم عقل
إشراف: الدكتور مجد الجمالي
Predictive modeling of Zinc concentration in Arabidopsis halleri leaves using Artificial Intelligence and plant photos
This study aimed firstly, to develop a logistic regression model to predict Zn contamination in soil using a combination of morphological and physiological features extracted through traditional measurement methods from 1000 individuals of A.halleri, as well as plant color (chlorophyll content) features extracted from plant images using machine learning approaches. Secondly, To develop a multiple machine learning model to predict Zn concentration in A.halleri leaves based on the same plant features. The logistic regression model demonstrated a remarkable training accuracy of 0.9423 when using the traditional features alone, indicating the model's exceptional ability to classify the training data into polluted and non-polluted conditions. Interestingly, when the plant color feature derived from images was incorporated, the training accuracy increased to 0.954, and the test accuracy improved from 0.905 to 0.94, confirming the importance of the image-based feature in enhancing the model's performance. This study contributes to the growing body of evidence that emphasizes the significance of evaluating machine learning models not only on their training performance but also on their ability to generalize to new, unseen data. The developed model has the potential to assist in the prediction of Zn contamination in soil, leveraging both traditional plant features and image-based features, which can substantially reduce the cost and time requirements associated with traditional laboratory analyses
إعداد: الطالبة ديما سليمان
إشراف: الدكتور ياسر خضرا
The Role of Bioinformatics in Determining the Genes Responsible for Certain Hereditary Dental Diseases
The role of bioinformatics in identifying genes responsible for hereditary dental diseases is of critical importance. This study aims to explore genetic components associated with such conditions by leveraging advanced bioinformatics tools. By analyzing genetic sequences and biological pathways related to enamel formation and dental caries, the research seeks to enhance our understanding of the genetic factors influencing these dental conditions. Data were collected from publicly available genetic databases and reviewed through relevant literature. Utilizing bioinformatics tools like BLAST and Clustal Omega, the study focused on sequence alignment, multiple sequence alignment, and pathway analysis. While specific mutations were not identified, the analysis provided valuable insights into genetic sequences and their potential impacts on protein structure and function. The findings underscore the essential role of bioinformatics in dental research, emphasizing the need for interdisciplinary approaches. This study contributes to the advancement of dental genetics and proposes recommendations for future research to further investigate and develop effective diagnostic and therapeutic strategies.
إعداد: الطالبة كارلا ماجد زياده
إشراف: الدكتور ينال قدسي