This thesis presents an open-source, modular Python pipeline for the automated extraction, and FAIR-compliant packaging of sequencing metadata at the Laboratory of Genomics and Epigenomics (LAGE), Area Science Park, Trieste. The facility operates three major sequencing platforms (Illumina NovaSeq6000, Oxford Nanopore PromethION 24, Illumina iScan) generating approximately 100 TB of data per year across heterogeneous file formats with no shared metadata schema.
The pipeline employs content-based file detection through eleven specialised extractor modules to produce JSON outputs structured around a common top-level schema aligned with the DECOS institutional Common Data Model. Dedicated utility modules reconstruct sample lineage across runs and sequencing targets through flexible identifier matching. FAIR packaging modules assemble datasets and metadata as RO-Crate 1.2 research objects serialised in JSON-LD.
Evaluated against three real-world datasets, the pipeline achieved zero misclassifications and all generated RO-Crates were validated against the RO-Crate 1.2 specification. The pipeline eliminates manual metadata curation, enables sample-level provenance reconstruction, and produces FAIR-compliant research objects ready for institutional deposit.