Scalable Genome Annotation at CEA-Genoscope
From March 10 to May 2, 2025, Fabio Zanarello, LongTREC Doctoral Candidate at the Centre for Genomic Regulation (CRG), had the opportunity to carry out a research secondment at CEA-Genoscope, supervised by France Denoeud and Jean-Marc Aury. This secondment offered valuable growth, both in scientific expertise and personal development.
My goal? Deepening expertise in scalable genome annotation for large-scale biodiversity genomics projects. This covered integrating long-read RNA-seq evidence into ab initio gene prediction, and developing a module to retrieve and summarise gene annotations from phylogenetically related species.
Fabio reflects on his experience:
What I Learned ๐
I tested how best to incorporate gene models from IsoQuant into the GeneID ab initio gene predictor, focusing on the effect of combining intron- and exon-level hints. The resulting approach improved model sensitivity.
Takeaways ๐ฑ
I developed phylocontext, a module built on the NCBI Datasets API that retrieves annotated genomes from closely related species and extracts summary statistics such as gene counts, exon and intron lengths, and feature distributions. It gives reference expectations for benchmarking annotations of newly sequenced organisms.
Impact ๐
The gene prediction methodology served as foundational material for the LongTREC Summer School training sessions.
Overall, this secondment delivered a tested annotation strategy and a reusable tool for comparative annotation.
