Questions: Gene Prediction and Annotation

Question 1

Why is gene prediction significantly more difficult in eukaryotic genomes compared to prokaryotic genomes?

• A. Eukaryotic genomes are always larger
• B. Eukaryotic genes contain introns that interrupt coding sequences, requiring splice site prediction
• C. Eukaryotic organisms have more genes
• D. Eukaryotic DNA uses different base pairs

Answer: (B)

The primary difficulty is introns. Prokaryotic genes are contiguous — an ORF search reliably identifies them. Eukaryotic genes are split into exons and introns, so the coding sequence is fragmented across the genomic DNA. The gene predictor must identify each exon boundary (splice donor and acceptor sites), assemble the correct combination of exons, and distinguish real splice sites from the many sequences that resemble them. Alternative splicing adds further complexity.

Question 2

An ab initio gene predictor uses only the DNA sequence and statistical models to predict genes, without any experimental evidence.

• T. True
• F. False

Answer: True

Ab initio (from the beginning) gene predictors rely entirely on sequence features and statistical models — typically hidden Markov models trained on known genes from the same or related species. They recognize patterns like codon usage bias, splice site consensus sequences, promoter elements, and polyadenylation signals. While powerful, they are less accurate than evidence-based methods that incorporate RNA-seq data or protein alignments, which is why modern pipelines combine both approaches.

Question 3

Explain why RNA-seq data is particularly valuable for refining eukaryotic gene predictions.

Model answer: RNA-seq captures the actual transcripts produced by cells, directly revealing which regions of the genome are transcribed and how exons are spliced together. Reads that span exon-exon junctions provide definitive evidence for intron locations and alternative splicing patterns. This experimental evidence resolves ambiguities that ab initio methods cannot — such as which of several possible splice sites is actually used, whether a predicted gene is truly expressed, and the full diversity of transcript isoforms.

Before RNA-seq, gene annotation relied heavily on ESTs (expressed sequence tags) and cDNA libraries, which had limited coverage and tissue representation. RNA-seq provides comprehensive, quantitative, tissue-specific transcript evidence that has dramatically improved annotation quality in every genome where it has been applied.