Assembly and gene_prediction
- 1. Genome assembly and gene prediction Robin Ohm Microbiology Department 1 November 2017 Master Course Introduction to Bioinformatics
- 2. Genome sequencing: pipeline Organism DNA extraction DNA Sequencing Sequence reads (100-300 bp) Assembly Gene prediction
- 4. Making contigs Gap Single stranded • Assemble the short sequences into ‘contigs’: contiguous stretches of sequence • Overlapping short sequences are merged into a consensus sequence
- 6. Contig assembly: CAP3 Algorithm • Uses quality values to define 5’ and 3’ clipping regions • Assembled into contig if both reads have high similarity sequences are high quality • Quality values determine which base in consensus
- 8. Misassembly caused by pseudo-overlap
- 9. Misassembly caused by pseudo-overlap
- 10. Usefulness of paired end-sequences
- 11. Paired-end reads are very useful • Improving the assembly (linking contigs into scaffolds) • Detecting genome changes relative to reference
- 13. Step 1: masking repetitive regions Repetitive regions: DNA sequences that occur many times in the genome • Transposons / Transposable Elements – Self-replicating sequences – Usually of viral origins – Contains genes (or at least Open Reading Frames) • Low-complexity regions. For example: – (G)n : GGGGGGGGGGGGGGGGGGGGGGGGGGGG – (GAT)n : GATGATGATGATGATGATGATGATGATGATG
- 14. Repeat masking
- 15. Step 2. Predict genes • Prokaryotes – Scan for Open Reading Frames (ORFs) • Eukaryotes – Simple ORF scanning not possible due to introns
- 16. Prokaryotes: scan for open reading frames
- 17. Eukaryotes: Introns complicate ORF scanning Recognition of intron splice sites is difficult
- 18. Gene prediction in eukaryotes 3 strategies • Ab initio prediction: uses statistical parameters (‘rules’) to find exons, introns, terminators, promoters, etc. Not evidence- based. • Homology-based: uses known genes from related organisms to find genes • Transcript-based: uses expression data (cDNA) to find genes
- 19. Ab initio gene prediction • Not evidence-based • Parameter for: – Start and Stop codons – Length of open reading frame (ORF) – Ribosome binding site or Kozak sequence – Codon usage – Bias towards G or C as third base – Average GC content (CpG islands) – TATA box – Polyadenylation sites (consensus AATAAA) – 5’ and 3’ Splice signals – Hexamer base composition (discrimination between codon and non- coding) • These parameters are organism-specific. Many differences between e.g. animals, plants, fungi • Train the predictor using a subset of known genes
- 20. Homology-based gene prediction • Uses genes (proteins) from other organisms to find genes • Works for well-conserved genes Source: http://bioinformatica.upf.edu
- 21. Transcript-based gene prediction • Uses cDNA to identify genomics regions that are expressed • ESTs: expressed sequence tags
- 22. Considerations mRNA analysis • Open reading frames – Mature mRNA contains ORF – All internal exons contain open “read-through” – Pre-start and post-stop sequences are UTRs • Posttranscriptional modification – 5’-CAP, polyA tail, splicing • Multiple translates – One gene – many proteins via alternative splicing
- 26. Overview gene predictors • Fungal genome: Schizophyllum commune Various gene predictors ESTs (mRNA) Homology with other organisms (BLAST)
- 27. State of the art: BRAKER1 • Ab initio gene predictor, based on Augustus • Can be trained using RNA-seq data • First it learns the rules from experimental data • Then applies those rules to the entire genome Introns Exons
- 28. Which gene prediction is correct? • Gene predictors routinely make different predictions • Manual curation is essential • Based on evidence – RNA-seq expression data – Homology
- 29. Correctly or incorrectly predicted?
- 30. Common problem in fungal genomes • Fungal genomes have very high gene density • Neighboring genes are sometimes fused into one gene • Chimera
- 31. Functional annotation • Next step: what do these predicted genes do? • Homology-based – Known gene from another (related?) organism • Presence of conserved protein domains – PFAM domains: DNA binding domain, kinase, etc • Transcript analysis – Differentially regulated? – RNA-Seq • Proteomics
- 32. Functional annotation • PFAM domains (protein family)