Data mining of molecular data resulting from environmental exposure to xenobiotics

Microplastic exposure is increasingly recognized as a stressor for human tissues, yet its molecular impact across regulatory layers remains incompletely defined. This thesis characterizes the multi-layer transcriptomic response to polystyrene in airway (BEAS-2B) and liver (HepG2) models across dose (10, 100 µg/ml) and time (24, 48, 72 h), and delivers a portable, auditable analysis workflow. An end-to-end Snakemake pipeline performs acquisition, quality control, trimming, alignment (HISAT2 for long-RNA; a Bowtie1-based, contaminant-depletion then miRNA alignment for small-RNA), differential analysis (DESeq2 with effect-size shrinkage), isoform switch detection (IsoformSwitchAnalyzeR), validated miRNA-target integration, and functional enrichment (clusterProfiler/ReactomePA, ORA and GSEA). Reporting is automated to produce synchronized HTML summaries with full provenance. Biologically, polystyrene elicits a dose- and time-structured remodeling of expression programs. Thousands of differentially expressed transcripts and ~1.6k differentially expressed genes are accompanied by extensive isoform reprogramming (>3,000 significant switches across ~2,300 genes) with predicted consequences that alter coding capacity, intron retention, and nonsense-mediated decay susceptibility. Although significant miRNA changes are concentrated in a subset of high-dose contrasts, the associated validated target networks are large, indicating substantial potential for post-transcriptional control. Enrichment analyses converge on mitochondrial respiration/oxidative phosphorylation, translation and mRNA surveillance (including NMD), stress-activated signaling (MAPK/NF-κB), and p53-aligned checkpoint and DNA-repair programs, with detoxification modules appearing in a context-dependent manner. Cell-type differences align with physiology: BEAS-2B emphasizes innate/epithelial stress programs, whereas HepG2 shows pronounced metabolic, mitochondrial, and translational signatures. Numerous novel isoforms contribute to differential expression and switching, underscoring transcriptome plasticity beyond current annotation. Methodologically, the workflow enforces determinism and comparability: identical code paths, centralized thresholds, uniform backgrounds for enrichment, explicit handling of “empty” branches, and environment-portable resource configuration. Every reported artifact is bound to a file target, enabling re-execution under updated references or parameters and facilitating external audit. Together, the results indicate that polystyrene exposure is not neutral for the tested cell types; rather, it induces coordinated transcriptional, splicing, and miRNA-mediated responses along axes plausibly linked to pathology under sustained or mixed exposures. The pipeline generalizes to other materials and datasets, providing a reusable platform for multi-layer toxicogenomic investigation.