Profile
MSc Bioinformatics student at the University of Bologna with a background in Medical Laboratory Sciences and a focus on computational protein biology, protein language models, and machine-learning-based protein sequence analysis. Experienced with ESM-2 representations, sequence-function prediction, HMMER/MMseqs2 workflows, and reproducible machine-learning pipelines. Interested in developing learning-based methods for protein sequence, structure, function, and molecular interactions.
Research Interests
- Protein language models and representation learning
- Protein sequence–structure–function modelling
- Protein–protein and protein–peptide interaction modelling
- Generative and structure-aware protein design
- Remote homology detection and functional annotation
Education
University of Bologna — MSc in Bioinformatics
Oct 2024 – Present
Coursework: sequence alignment, RNA-seq, microarrays, statistical analysis in R; machine learning and deep learning with scikit-learn and TensorFlow; SQL/graph queries, Docker, HPC/Slurm.
Iran University of Medical Sciences — BSc in Medical Laboratory Sciences
Sept 2019 – Jul 2023
Relevant coursework: Molecular Diagnostics, Immunoassays, Clinical Chemistry, Haematology, Microbiology, Virology.
Publications
Farmani, S., Dehghanbanadaki, N., Balooei, M.(equal-contributing author), Mehralitabar, H., Sabaghian, H., & Naderi-Manesh, H. (2025). Mechanism-based targeting of lactate dehydrogenase tetramerization process through natural derived compounds: A survey of molecular docking, molecular dynamics, and QSAR analysis.
Computers in Biology and Medicine, 197, 111000. Paper →
Screened approximately 5,000 naturally derived compounds against four LDH tetramerization interfaces, evaluating top candidates using molecular docking, 100-ns molecular dynamics simulations, MM-PBSA, PCA/DCCM, pharmacophore, ADMET, and QSAR analyses. Identified candidate inhibitors with distinct effects on LDH conformational dynamics, highlighting two compounds as promising modulators of the A2 and B2 tetramerization interfaces respectively.
Research Experience
Protein Language Model-Based Signal Peptide Prediction
Independently implemented MSc Bioinformatics project · Sep 2025 – Nov 2025
Built an end-to-end protein sequence modelling pipeline for eukaryotic signal-peptide prediction using high-confidence UniProtKB/Swiss-Prot annotations. Applied MMseqs2 clustering for redundancy reduction and constructed biologically meaningful train/benchmark splits. Compared von Heijne PSWM, biochemical-feature SVM, CNN-BiLSTM, and ESM-2 embedding-based models, achieving F1 = 0.962 and MCC = 0.957 on a held-out benchmark with the ESM-2-based model. Analyzed amino-acid composition, signal-peptide lengths, kingdom-specific bias, and cleavage-site motifs for biological interpretation. Code
Protein Domain and Functional Annotation Using HMMER and MMseqs2
Jun 2025 – Jul 2025
Built an automated protein domain annotation workflow using profile HMMs, HMMER, and MMseqs2 for large-scale sequence classification. Processed protein structure-derived alignments to support domain-level annotation and sequence-based functional inference. Compared profile-based and sequence-similarity-based classification workflows, achieving MCC = 0.997 and MCC = 0.991 respectively, connecting classical homology and profile-HMM approaches to modern protein representation learning tasks. Code
Computational Epigenomics Analysis of DNA Methylation (Illumina 450K)
Jun 2025 – Jul 2025
Conducted an R/Bioconductor workflow for Illumina 450K DNA methylation array analysis, including quality control, normalization, PCA-based batch-effect assessment, and differential methylation analysis. Identified differentially methylated regions and visualized methylation patterns using heatmaps, volcano plots, and exploratory statistical summaries. Code
Chest X-Ray Pneumonia Detection via Transfer Learning
Mar 2026 – Apr 2026
Compared ResNet50, EfficientNetB0, and DenseNet-121 XRV for pneumonia classification. Addressed class imbalance and evaluated threshold-selection strategies using clinically relevant metrics, applying ensembling, test-time augmentation, and Grad-CAM-based interpretability. DenseNet-121 XRV achieved the strongest performance under the maximum-MCC threshold strategy. Code
Skills
Protein Bioinformatics & Sequence Analysis: ESM-2 protein language model embeddings, protein representation learning, signal peptide prediction, sequence-function prediction, protein-domain annotation, HMMER, MMseqs2, BLAST, Clustal, UniProt, PDB
Machine Learning & Deep Learning: PyTorch, scikit-learn, TensorFlow, SVMs, CNN-BiLSTM, transfer learning, representation learning, model evaluation, threshold optimization, class-imbalance handling, ensembling, Grad-CAM-based interpretability
Functional Genomics & Epigenomics: DNA methylation analysis (Illumina 450K arrays, minfi), differential methylation region (DMR) detection, PCA-based batch-effect assessment, R/Bioconductor
Reproducible Workflows & Tools: Python, R, Bash/Linux, SQL, Git/GitHub, Docker, HPC/Slurm, R Markdown, Jupyter Notebooks, conda environments
Work Experience
Clinical Laboratory Internships — Tehran, 2022–2023
Rotations in haematology, microbiology, clinical chemistry, and biochemistry across hospital and diagnostic laboratory settings.
Languages
English: Proficient (TOEFL iBT 93, CEFR B2) · Italian: Basic communication · Persian (Farsi): Native
Contact
- Email:mahan.balooei@studio.unibo.it·mahanbalooei2@gmail.com
- GitHub:MahanBalooei
- LinkedIn:in/mahan-balooei
- ORCID:0009-0006-5358-0784