Tomato genome annotation pipeline in Cyrille2 Erwin Datema
Contents of the annotation pipeline Annotation on the BAC level Gene prediction Repeat identification Other features Annotation on the gene level (work in progress) blastx vs NCBI’s nr (sequence similarity) InterProScan (domain identifcation)
Ab initio gene structure prediction Ab initio predictors included in the pipeline Genscan GlimmerHMM (trained on tomato!) GeneId (has been trained on Solanaceae) SNAP Augustus (predicts alternative spliced variants)
Alignment-based gene structure prediction (1) Transcript alignment (blastn + Sim4) SGN tomato UniGenes (34.829 UniGenes) SGN potato UniGenes (31.072 UniGenes) SGN coffee UniGenes (13.171 UniGenes) SGN pepper UniGenes (9.554 UniGenes) SGN petunia Unigenes (5.135 UniGenes) SGN S. melongena UniGenes (1.841 UniGenes) NCBI full-length tomato cDNAs (678 cDNAs) Protein alignment (tblastn + GeneWise) TAIR6 Arabidopsis thaliana proteome (30.690 proteins) TIGR4 Oryza sativa proteome (62.827 proteins) UniProt Plant division (17.831 proteins)
Additional feature prediction Repeat Identification Tandem Repeats Finder RepeatMasker RepBase + ‘default’ features (low complexity, etc) TIGR Solanum lycopersicon repeat library V2 SGN Solanum lycopersicon UniRepeats Feature prediction tRNAscan-SE MarScan GeneSplicer Marker identification (blastn + Sim4)
Preliminary results Annotation of chromosome 6 BACs phase 1, 2 and 3 632 contigs Older version of the pipeline GlimmerHMM only trained on Arabidopsis 2 UniGene sets (tomato, potato) 2 protein sets (Arabidopsis, UniProt plant) Protein alignment parameters too strict
The genomic landscape of chromosome 6 632 contigs have been annotated Length of contigs varies between 348 – 148.256 nt Average length of 9.061 nt, median length of 5.105 nt Total length of 5.726.791 nt GC content: 29.9% min, 34.1% avg, 42.2% max (sequences longer than 10.000 nt)
Ab initio gene prediction Note: Augustus predictions include up to 3 splice variants per gene Estimated gene density is 1 gene per 5 kb ~1.200 genes in currently sequenced BACs
Transcript alignment-based gene prediction Tomato 34.829 UniGenes (derived from 239.593 ESTs) 574 hits to the contigs Potato 31.072 UniGenes (derived from 133.657 ESTs) 631 hits to the contigs
Protein alignment-based gene prediction UniProt Plant proteins 17.378 protein sequences from the plant kingdom 195 hits to the contigs Arabidopsis thaliana TAIR6 annotation 30.690 protein sequences 228 hits to the contigs
Repeat density TIGR Tomato Repeat Library (95 repeats) 118 regions spanning 53.024 nt Minimum 48 nt, average 449 nt, maximum 7.675 nt SGN Tomato UniRepeats (668 repeats) 2.860 regions spanning 1.220.101 nt Minimum 10 nt, average 427 nt, maximum 8.896 nt Tandem repeats 1.313 regions spanning 157.921 nt Minimum 24 nt, average 120 nt, maximum 2.526 nt
Additional features 74 markers could be aligned alignment quality unverified 39 predicted tRNA genes 1.301 predicted MAR/SAR elements
Generic Genome Browser (1)
Generic Genome Browser (2)
Generic Genome Browser (3)
Recent work GeneModelCollector JIGSAW Tries to find ‘full’ open reading frames in aligned UniGenes Automatic generation of gene predictor training set Parameters? JIGSAW Appears not to provide a prediction for every region which contains annotations Training?
Future Work – Tomato Annotation Pipeline Gene prediction Combining predictions into a single consensus model Train individual predictors with recently curated tomato gene set Automated functional annotation of genes “Giving a biological meaning to the nicely colored bars” blastx InterProScan
Future Work – Tomato Genome Browser Annotation of features Meaningful names for features such as genes, marker alignments, blast hits More detailed and better readable data when clicking on a feature Links to external data sources NCBI GenBank SGN
Acknowledgements Cyrille2 development Mark Fiers Ate van der Burgt Joost de Groot Tomato BAC sequencing (chromosome 6) Greenomics Supervision Willem Stiekema Roeland van Ham