1. Tag profiling is dead... October 2009 Claudia Voelckel Patrick Biggs...long live mRNA-Seq!

2. Expression Studies in the New Zealand Flora Species diversification & local adaptation Hybridization & polyploidy Biological processes that differ between species and populations Adaptive gene sets We are interested in: Ranunculus Ourisia Hebe PachycladonNothofagus Totara

3. 3 DNA chip with gene probes AAAAAA3’ TTTTTT5’ green-labeled cDNA AAAAAA3’ TTTTTT5’ red-labeled cDNA Expression Studies the Familiar Way: Microarrays Sample 1 AAAAAA3’ mRNA Sample 2 AAAAAA3’ mRNA DATA ANALYSIS intensity 1 intensity 2 Expression ratio:log

4. 4 Expression Studies Revolutionized: Tag Profiling If needed – build reference transcriptome through RNA seq 1212 2121 1111 count 1 count 2 log STATISTICAL ANALYSIS Solexa Genome Analyzer Sample 1mRNA AAA3’ Sample 2mRNA AAA3’ 18 bp tag library AAA3’ 18 bp tag library AAA3’ Sample 1 Sample 2 Reference TAG MAPPING

5. Advantages & Challenges of Tag Profiling  open to any organism (with a reference transcriptome)  any expressed transcript detectable (1 copy/cell)  less RNA needed (tag profiling = 1µg, microarrays = 100 µg)  minor data normalization, cross-species comparisons easier Advantages Challenges  mapping 18 bp tags (sequence differences Pachycladon/Arabidopsis)  counting tags per gene (noise, location, abundance)  statistical analysis of differential expression (proportion data)

6. Tag Profiling Guinea Pig: Previous Microarray Study HabitatRosette Flowering Fruiting HabitatRosette Flowering Fruiting Pachycladon fastigiataPachycladon enysiivs. Voelckel et al. 2008, Molecular Ecology, 17: 4740–4753 Comparative gene expression study using Arabidopsis microarrays

7. P. enysiiP. fastigiata Probability of differential expression ( log odds ratio) Magnitude of differential expression (log fold change) ESM1 ESP  Arabidopsis microarray (20,468 genes)  310 genes (1.5%) up in P. fastigiata 324 genes (1.6%) up in P. enysii  up-regulation of ESP and ESM1 predict P. fastigiata to produce isothiocyanates and P. enysii to produce nitriles  prediction confirmed by HPLC  role for herbivory in species diversification? Microarray Study Results

8. Tag Profiling Results  8 data sets from different mapping strategies (ELAND, MySQL)  each analyzed with different normalization parameters (R, edgeR)  results vary! Example:  data set 2: 17423 A. thaliana loci noise filter 10 count most abundant tag per gene  analyzed with tagwise normalization -log2(1.5) < log fold ratio < log2 (1.5)  2654 genes (15.2%) up in P. fastigiata 1857 genes (10.7%) up in P. enysii P. enysiiP. fastigiata

9. Microarrays (MA) vs. Tag Profiling (TP)  more differentially expressed genes in TP (10.7-15.2% ) than with MA (1.5-1.6% ) MA: 20,468 genes 310 up in PF 324 up in PE TP: 17,423 genes 2654 up in PF 1857 up in PE  biological inferences from both studies identical PF MA TP 41269 2613 PE 502741807 MA TP  13.2% (PF) and 15.4 % (PE) of MA results confirmed by TP results

10. “...not a popular product, too expensive, tricky chemistry.. instead use: mRNA-Seq!” Tag Profiling is dead, long live mRNA-Seq! 2 Oct 09: “Illumina is discontinuing the support of Tag Profiling and will no longer be manufacturing the reagent kits for this application.”  One year later: Tag profiling works for a non-model plant with a distant reference transcriptome! Let’s do more experiments!

11. 11 Expression Studies Revisited: mRNA-Seq If needed – build reference transcriptome through RNA seq Sample 1 AAA3’ Sample 2mRNA Solexa Genome Analyzer AAA3’ cDNA library Sample 1 Sample 2 Reference READ MAPPING 1212 2121 1111 count 1 count 2 log STATISTICAL ANALYSIS gene length

12.  new mapping strategies needed  different statistical treatment required  hardly any R packages available yet Advantages & Challenges of mRNA-Seq Advantages Challenges  whole transcriptome coverage  longer reads reduce mapping noise and unmapped reads  multiplex-compatible  adequate coverage (too high with tag profiling)  additional benefits: EST libraries, SNPs  disentangling expression of allopolyloid copies may be easier

13. Ranunculus OurisiaPachycladon Experiences with mRNA-Seq Tuatara mRNA-Seq runs (75bp paired end) so far: await assembly and analysis EST data base built for P. fastigiata analysis in progress

14. Straight from the Pachycladon EST library: Evidence for allopolyploid copies (e.g. glucosinolate hydrolysis gene)

15. THANKS TO: Genome Service Patrick Biggs Lorraine Berry Lesley Collins, Maurice Collins, Pete Lockhart Helene Kretzmer Marsden Alexander von Humboldt Foundation