1. 最佳的多重序列比對方法針對基因組 領域 Cédric Notredame Comparative Bioinformatics Group Bioinformatics and Genomics Program

2. Getting the best out of multiple sequence alignment methods in the genomic era Cédric Notredame Comparative Bioinformatics Group Bioinformatics and Genomics Program

3. Which Tool for Which Sequence ?

4. In- SilVo Biology In Silico Biology – Making Sense of digital data In Vivo Biology – Recording data in a living Cell In SilVo Biology – Connect In-Vitro and In-Vivo In-Vivo: High-throughput recording In-Silico: High-Throughput analysis

6. Is it Possible to Compare all Types of Sequences ? Non Transcribed World – Genes/Full Genomes Lagan, TBA – Promoter Regions Meta-Aligner Motifs Finders – Nucleosome ??? Multiple Genome Aligners – Not Very Accurate – Very Fast – Deal with rearrangements

7. Multiple Genome Alignments and re-sequencing Before – Re-sequence Human Genomes – Map the Reads onto the reference genome Now – Re-sequence – Assemble – Align – Non trivial with very large datasets

8. Is it Possible to Compare all Types of Sequences ? RNA Comparison – Less Accurate than Proteins – Secondary Structures ncRNA World – Sankoff Time O(L 2n ) Space O(L 3n ) – Consan – R-Coffee

9. Is it Possible to Compare all Types of Sequences ? Protein Comparisons – Very Accurate – 3D-Structure Improves it Protein Aligners – ClustalW – T-Coffee – 3D-Coffee

10. What Changes with 1000 Genomes?

11. Phylogeny Vs Function Function – Low level => Biochemistry => Protein Domains – High Level => Metabolic Pathway => Orthology Orthology – Phylogenetic Analysis – Phylogenetic Analysis =>Accurate Alignments

12. Duplication node Speciation node or leaf one2one one2many many2many apparent one2one (Adpated from “Going beyond AGC and T, E. Birney)

13. Using The tree Correct Tree  Correct Orthologous Assignment  Correct Functional Prediction

14. The Alignment that Hides The Forest…

15. Phylogenetic Trees and Multiple Sequence Alignments

17. Genomic Era: The Goal 10.000 Sequences: interspecies 1 Billion: Re-sequencing Incorporation of ALL experimental Data – Structure, Genomic, ChIp-Chip, ChIp-Seq… Alignments suitable for all applications of comparative genomics – Homology Modeling (function) – Functional Analysis – Phylogenetic Reconstruction – 3D-Modelling Accurate Alignments for ALL kind of data Non Transcribed DNA Transcribed DNA Translated DNA

18. Genomic Era Challenges Accuracy – Proteins: 30% is the limit – DNA/RNA 70% is the limit Scale – With too many sequences algorithms lose in accuracy Data Integration – Structure – Homology – Genomic Structure – Function – Proteomics Methods – Wealth of alternative methods – Poorly Characterized

19. Consistency and Data Integration Most methods rely on the progressive algorithm Consistency based methods have been designed as an extension Consistency based alignment methods have been designed to: – Better extract the signal contained in the data – Integrate/Confront existing methods – Integrate/Confront heterogeneous types of Information

20. The Progressive Alignment Algorithm

21. T-Coffee and Concistency… SeqA GARFIELD THE LAST FAT CAT SeqB GARFIELD THE FAST CAT SeqC GARFIELD THE VERY FAST CAT SeqD THE FAT CAT SeqA GARFIELD THE LAST FA-T CAT SeqB GARFIELD THE FAST CA-T --- SeqC GARFIELD THE VERY FAST CAT SeqD -------- THE ---- FA-T CAT

22. T-Coffee and Concistency… SeqA GARFIELD THE LAST FAT CAT Prim. Weight =88 SeqB GARFIELD THE FAST CAT --- SeqA GARFIELD THE LAST FA-T CAT Prim. Weight =77 SeqC GARFIELD THE VERY FAST CAT SeqA GARFIELD THE LAST FAT CAT Prim. Weight =100 SeqD -------- THE ---- FAT CAT SeqB GARFIELD THE ---- FAST CAT Prim. Weight =100 SeqC GARFIELD THE VERY FAST CAT SeqC GARFIELD THE VERY FAST CAT Prim. Weight =100 SeqD -------- THE ---- FA-T CAT

23. T-Coffee and Concistency… SeqA GARFIELD THE LAST FAT CAT Prim. Weight =88 SeqB GARFIELD THE FAST CAT --- SeqA GARFIELD THE LAST FA-T CAT Prim. Weight =77 SeqC GARFIELD THE VERY FAST CAT SeqA GARFIELD THE LAST FAT CAT Prim. Weight =100 SeqD -------- THE ---- FAT CAT SeqB GARFIELD THE ---- FAST CAT Prim. Weight =100 SeqC GARFIELD THE VERY FAST CAT SeqC GARFIELD THE VERY FAST CAT Prim. Weight =100 SeqD -------- THE ---- FA-T CAT SeqA GARFIELD THE LAST FAT CAT Weight =88 SeqB GARFIELD THE FAST CAT --- SeqA GARFIELD THE LAST FA-T CAT Weight =77 SeqC GARFIELD THE VERY FAST CAT SeqB GARFIELD THE ---- FAST CAT SeqA GARFIELD THE LAST FA-T CAT Weight =100 SeqD -------- THE ---- FA-T CAT SeqB GARFIELD THE ---- FAST CAT

24. T-Coffee and Concistency… SeqA GARFIELD THE LAST FAT CAT Weight =88 SeqB GARFIELD THE FAST CAT --- SeqA GARFIELD THE LAST FA-T CAT Weight =77 SeqC GARFIELD THE VERY FAST CAT SeqB GARFIELD THE ---- FAST CAT SeqA GARFIELD THE LAST FA-T CAT Weight =100 SeqD -------- THE ---- FA-T CAT SeqB GARFIELD THE ---- FAST CAT

25. T-Coffee and Concistency…

27. Methods Data Scalability

28. A Brief History of Consistency A Long Chain of Small Contributions…

29. Consistency Based Algorithms Gotoh (1990) – Iterative strategy using consistency Martin Vingron (1991) – Dot Matrices Multiplications – Accurate but too stringeant Dialign (1996, Morgenstern) – Concistency – Agglomerative Assembly T-Coffee (2000, Notredame) – Concistency – Progressive algorithm ProbCons (2004, Do) – T-Coffee with a Bayesian Treatment – Biphasic Gap Penalty AMAP (Schwarz, 2007) – ProbCons Consistency – Replace Progressive alignment with simulated Annealing – Hard to distinguish from ProbCons FSA ( Patcher, 2009) – AMAP with automated parameter estimation – Hard to distinguish from ProbCons

30. Choosing the right modeling method M-Coffee

31. Combining Many MSAs into ONE MUSCLE MAFFT ClustalW ??????? T-Coffee

32. Consistency and Accuracy

33. Integrating New Types of Data Template Based Sequence Alignments

34. Experimental Data … TARGET Experimental Data … TARGET Template Aligner Template-Sequence Alignment Primary Library Template Alignment Template based Alignment of the Sequences Templates TARGET

35. Exploring The Template World TemplateGeneratorAlignment Method RNA StructurePredictionRNA Aligner Protein StructureBLAST vs PDB3D Aligner ProfileBLAST vs NRProfile/Profile Alignment Gene StructureENSEMBLGenome Aligner PromoterTransfacMeta-Aligner

36. Exploring The Template World TemplateGeneratorAlignment Method Mode RNA Structure PredictionRNA Aligner R-Coffee Protein Structure BLAST /PDB3D Aligner 3D-Coffee Profile BLAST/NRProfile/Profile PSI-Coffee Gene Structure ENSEMBLGenome Aligner Exoset Promoter TransfacMeta-Aligner Meta-Coffee

37. 3D-Coffee/Expresso Incorporating Structural Information

38. Expresso: Finding the Right Structure Sources Templates Library BLAST SAP Template Alignment Source Template Alignment Remove Templates Templates

39. PSI-Coffee Homology Extension

40. Exploring The Template World

41. What is Homology Extension ? LL L ? -Simple scoring schemes result in alignment ambiguities

42. What is Homology Extension ? LL L L L L L L L L L I V I L L L L L L L Profile 1 Profile 2

43. What is Homology Extension ? LL L L L L L L L L L I V I L L L L L L L Profile 1 Profile 2

44. PSI-Coffee: Homology Extension Sources Templates Library BLAST Template Alignment Source Template Alignment Remove Templates Templates Profile Aligner

45. Benchmarks

46. Method TemplateScoreComment ClustalW-2ProgressiveNO22.74 PRANKGapNO26.18Science2008 MAFFTIterativeNO26.18 MuscleIterativeNO31.37 ProbConsConsistencyNO40.80 ProbConsMonoPhasicNO37.53 T-CoffeeConsistencyNO42.30 M-Coffe4ConsistencyNO43.60 PSI-CoffeeConsistencyProfile53.71 PROMALConsistencyProfile55.08 PROMAL-3DConsistencyPDB57.60 3D-CoffeeConsistencyPDB61.00Expresso Score: fraction of correct columns when compared with a structure based reference (BB11 of BaliBase).

47. Method TemplateScoreComment ClustalW-2ProgressiveNO22.74 PRANKGapNO26.18Science2008 MAFFTIterativeNO26.18 MuscleIterativeNO31.37 ProbConsConsistencyNO40.80 ProbConsMonoPhasicNO37.53 T-CoffeeConsistencyNO42.30 M-Coffe4ConsistencyNO43.60 PSI-CoffeeConsistencyProfile53.71 PROMALConsistencyProfile55.08 PROMAL-3DConsistencyPDB57.60 3D-CoffeeConsistencyPDB61.00Expresso Score: fraction of correct columns when compared with a structure based reference (BB11 of BaliBase). Consistency

48. Method TemplateScoreComment ClustalW-2ProgressiveNO22.74 PRANKGapNO26.18Science2008 MAFFTIterativeNO26.18 MuscleIterativeNO31.37 ProbConsConsistencyNO40.80 ProbConsMonoPhasicNO37.53 T-CoffeeConsistencyNO42.30 M-Coffe4ConsistencyNO43.60 PSI-CoffeeConsistencyProfile53.71 PROMALConsistencyProfile55.08 PROMAL-3DConsistencyPDB57.60 3D-CoffeeConsistencyPDB61.00Expresso Score: fraction of correct columns when compared with a structure based reference (BB11 of BaliBase). Homology Extension

49. Method TemplateScoreComment ClustalW-2ProgressiveNO22.74 PRANKGapNO26.18Science2008 MAFFTIterativeNO26.18 MuscleIterativeNO31.37 ProbConsConsistencyNO40.80 ProbConsMonoPhasicNO37.53 T-CoffeeConsistencyNO42.30 M-Coffe4ConsistencyNO43.60 PSI-CoffeeConsistencyProfile53.71 PROMALConsistencyProfile55.08 PROMAL-3DConsistencyPDB57.60 3D-CoffeeConsistencyPDB61.00Expresso Score: fraction of correct columns when compared with a structure based reference (BB11 of BaliBase). Structural Extension

50. T-Coffee and The World BLAST/ SOAP -Some Templates are obtained with a BLAST -Queries can be sent to the EBI or the NCBI -No Need for a Local BLAST installation Users sequences

51. Incorporating RNA Information Within the T-Coffee Algorithm

52. ncRNAs Can Evolve Rapidly CCAGGCAAGACGGGACGAGAGTTGCCTGG CCTCCGTTCAGAGGTGCATAGAACGGAGG **-------*--**---*-**------** GAACGGACCGAACGGACC CTTGCCTGGCTTGCCTGG G G A A CC A C G G A G A C G CTTGCCTCCCTTGCCTCC GAACGGAGGGAACGGAGG G G A A CC A C G G A G A C G

53. ncRNAs Can Evolve Rapidly CCAGGCAAGACGGGACGAGAGTTGCCTGG CCTCCGTTCAGAGGTGCATAGAACGGAGG **-------*--**---*-**------** CC--AGGCAAGACGGGACGAGAGTTGCCTGG CCTCCGTTCAGAGGTGCATAGAAC--GGAGG ** * *** * * *** ** Sequence Alignment (Maximizing Identity) -Incorrect -Not Predictive

54. The Holy Grail of RNA Comparison: Sankoff’ Algorithm

55. C C R-Coffee Extension G G TC Library G G Score X C C Score Y C C G G Goal: Embedding RNA Structures Within The T-Coffee Libraries The R-extension can be added on the top of any existing method.

56. R-Coffee + Structural Aligners MethodAvg BraliscoreNet Improv. direct +T+R +T+R ----------------------------------------------------------- Stemloc0.620.750.76 104113 Mlocarna0.660.690.71 101133 Murlet0.730.700.72 -132-73 Pmcomp0.730.730.73 142145 T-Lara0.740.740.69 -36 -8 Foldalign0.750.770.77 72 73 ----------------------------------------------------------- Dyalign---0.630.62 ------ Consan---0.790.79 ------ ----------------------------------------------------------- Improvement= # R-Coffee wins - # R-Coffee looses over 170 test sets

57. R-Coffee + Regular Aligners MethodAvg BraliscoreNet Improv. direct +T+R +T+R ----------------------------------------------------------- Poa0.620.650.70 48154 Pcma0.620.640.67 34120 Prrn0.640.610.66-63 45 ClustalW0.650.650.69 -7 83 Mafft_fftnts0.680.680.72 17 68 ProbConsRNA0.690.670.71-49 39 Muscle0.690.690.73-17 42 Mafft_ginsi0.700.680.72-49 39 ----------------------------------------------------------- Improvement= # R-Coffee wins - # R-Coffee looses over 388 test sets

58. Genomic Era Challenges Conclusion Template Based Alignments Meta-Methods M-Coffee Homology Extension (Proteins) R-Coffee Scaled Consistency

59. Open Questions Accurately Aligning non transcribed DNA Coping with One Billion Human Genomes

60. www.tcoffee.org cedric.notredame@crg.es Comparative Bioinformatics University College Dublin – Des Higgins – Orla O’Sullivan – Iain Wallace (UCD, IE) Berlin Free University – Knut Reinert – Tobias Rausch Swiss Intitute of Bioinformatics – Ioannis Xenarios – Sebastien Morreti Comparative Bioinformatics – Merixell Oliva – Giovanni Bussoti – Carsten Kemena – Emanuele Rainieri – Ionas Erb – Jia Ming Chang – Matthias Zytneki

61. www.tcoffee.org cedric.notredame@europe.com www.tcoffee.org

63. Why So Much Interest For Multiple Alignments ? Extrapolation Motifs/Patterns Phylogeny Profiles Structure Prediction SNP Analysis Reactivity Analysis Regulatory Elements

64. Phylogeny Vs Function: Applications Comparative Genomics => New Medium New Medium => New Clinical Test

65. Detecting ncRNAs in silico: a long way to go… RNAse P

66. Obtaining the Structure of a ncRNA is difficult Hard to Align The Sequences Without the Structure Hard to Predict the Structures Without an Alignment

67. R-Coffee: Modifying T-Coffee at the Right Place Incorporation of Secondary Structure information within the Library Two Extra Components for the T-Coffee Scoring Scheme – A new Library – A new Scoring Scheme

70. G-INS-i, H-INS-i and F-INS-i use pairwise alignment information when constructing a multiple alignment. The two options ([HF]-INS-i) incorporate local alignment information and do NOT USE FFT.

71. Molecular Biology Within the System Biology Era Protein A Interacts with Regulates Inhibits Protein B

72. Molecular Biology In the 1000 Genomes Era Protein A Interacts with Regulates Inhibits Protein B Variation Within Species: CNVs of A and SNPs Conservation Across Species

73. System Biology vs Comparative Genomics Systems Biology  Systems can be Understood Comparative Genomics  Systems can Evolve through Selection

74. Phylogeny Vs Function: Applications – Important  Application – Possible  Many New Genomes – Challenging  Too Many New Genomes

75. 3D-Coffee: Combining Sequences and Structures Within Multiple Sequence Alignments

76. Comparing Methods MAFFT

77. Some Benchmark: BB11 BaliBase BB11: 38 highly divergent (less than 25% id) datasets from BaliBase BB11: predicts 78% of the results measured on other datasets Blackshield, Higgins

78. PhD Fellowships www.crg.es

81. What ‘s in a Multiple Sequence Alignment Evolution Inertia Common Ancestry Shows up In the sequences Selection Important Features Are Preserved Functional Constraint Same Function Same Sequence Convergence Phylogenetic Footprint, Evolutionary Trace …

82. Which Tool for Which Sequence ?

83. Is it Possible to Compare all Types of Sequences Non Transcribed World – Genes/Full Genomes Lagan, TBA – Promoter Regions Meta-Aligner Motifs Finders – Nucleosome ??? Multiple Genome Aligners – Not Very Accurate – Very Fast – Deal with rearrangements

84. Is it Possible to Compare all Types of Sequences RNA Comparison – Less Accurate than Proteins – Secondary Structures ncRNA World – Consan – R-Coffee

85. Is it Possible to Compare all Types of Sequences Protein Comparisons – Very Accurate – 3D-Structure Improves it Protein Aligners – ClustalW – T-Coffee – 3D-Coffee

86. Why So Much Interest For Multiple Alignments ? Extrapolation Motifs/Patterns Phylogeny Profiles Structure Prediction SNP Analysis Reactivity Analysis Regulatory Elements

87. What’s in a Multiple Alignment ? The MSA contains what you put inside: – Structural Similarity – Evolutive Similarity – Sequence Similarity You can view your MSA as: – A record of evolution – A summary of a protein family – A collection of experiments made for you by Nature…

88. Producing The Right Alignment Multiple Sequence Alignments Influence Phylogenetic Trees Choice of Method is not Neutral – Different Methods – Different Alignments – Different Trees Using The Right Models insures Producing the right Tree

89. Model Based Alignments vs Naïve Alignments Naïve Alignment – Lexicographic Alignment – Maximizing the number of identities – At best using a substitution matrix Model Based Alignments – Using a model – Protein structure information – RNA Structure information – Combining/Confronting Modeling methods Template based Alignments – Model based Alignments through the use of Templates

90. T-Coffee and Model Based Alignments T-Coffee Algorithm Expresso: Aligning Protein Structures R-Coffee: Aligning RNA structures M-Coffee: Combining methods

91. T-Coffee and Concistency…

92. When Sequences Are not Enough 3D-Coffee and Expresso

93. 3D-Coffee: Combining Sequences and Structures Within Multiple Sequence Alignments

95. Where to Trust Your Alignments Most Methods Agree Most Methods Disagree

96. Conclusion Model Based Alignments Give the best Accuracy Template based alignment is a very efficient way to turn Naïve aligners into model based aligners Sequence Alignments are not necessarily reliable over their entire lengths

97. Manguel M, Samaniego F.J., Abraham Wald’s Work on Aircraft Suvivability, J. American Statistical Association. 79, 259-270, (1984)

98. Building and Using Models 35.67 Angstrom

99. Computing the Correct Alignment is a Complicated Problem

100. Stochastic Optimization

101. Exploration of Complex Optimization Problems With Multiple Constraints – Genomic Alignments – RNA Alignments Generation of Population of Suboptimal Solutions – Quality=f( optimality ) Specification of Concistency Objective Function of T- Coffee

102. Three Types of Algorithms Progressive: ClustalW Iterative: Muscle Concistency Based: T-Coffee and Probcons

103. T-Coffee and Concistency… Each Library Line is a Soft Constraint (a wish) You can’t satisfy them all You must satisfy as many as possible (The easy ones)

104. Concistency Based Algorithms: T-Coffee Gotoh (1990) – Iterative strategy using consistency Martin Vingron (1991) – Dot Matrices Multiplications – Accurate but too stringeant Dialign (1996, Morgenstern) – Concistency – Agglomerative Assembly T-Coffee (2000, Notredame) – Concistency – Progressive algorithm ProbCons (2004, Do) – T-Coffee with a Bayesian Treatment

105. How Good Is My Method ?

106. Structures Vs Sequences

107. T-Coffee Results Validation Using BaliBase

108. Too Many Methods for ONE Alignment M-Coffee

110. Estimating the Accuracy of your MSA

111. What To Do Without Structures

112. 3D-Coffee: Combining Sequences and Structures Within Multiple Sequence Alignments

113. Expresso: Finding the Right Structure Why Not Using Structure Based Alignments

114. Template Based Multiple Sequence Alignments

115. -Structure -Profile -… Sources Templates Library Template Aligner Template Alignment Source Template Alignment Remove Templates Templates -Structure -Profile -…

116. MethodScoreTemplatesPrefabHomstrad -------------------------------------------------------------- ClustalWMatrix----61.80---- KalignMatrix----63.00---- MUSCLEMatrix----68.0045.0 -------------------------------------------------------------- T-CoffeeConsistency ----69.9744.0 ProbConsConsistency----70.54---- MafftConsistency----72.20---- M-CoffeeConsistency----72.91---- MUMMALSConsistency----73.10---- -------------------------------------------------------------- Clustal-db MatrixProfiles-------- PRALINEMatrixProfiles----50.2 PROMALSConsistencyProfiles79.00---- SPEMMatrixProfiles77.00---- -------------------------------------------------------------- EXPRESSOConsistencyStructures----71.9 * T-LaraConsistencyStructures-------- -------------------------------------------------------------- Table 1. Summary of all the methods described in the review. Validation figures were compiled from several sources, and selected for the compatibility. Prefab refers to some validation made on Prefab Version 3. The HOMSTRAD validation was made on datasets having less than 30% identity. The source of each figure is indicated by a reference. *The EXPRESSO figure comes from a slightly more demanding subset of HOMSTRAD (HOM39) made of sequences less than 25% identical.

117. Improving The Evaluation

118. How Do We Perform In The Twilight Zone? Concistency Based Methods Have an Edge Hard to tell Methods Apart Sequence Alignment is NOT solved

119. More Than Structure based Alignments Structural Correctness Is Only the Easy Side of the Coin. In practice MSA are intermediate models used to generate other models: DataModel TypeBenchmark HomologyProfileYes EvolutionTreesNo Structure3D-StructureCASP FunctionAnnotationNo

120. Conclusion Template based Multiple Sequence Alignments Projecting any relevant information onto the sequences Using this Information Need for new evaluation procedures Functional Analysis Phylogenetic Analysis Homology Search (Profiles) Homology Modelling Integrating data  Making sure your bits of data can fight with one another

121. Turning Data into Models Data Columbus, considered that the landmass occupied 225°, leaving only 135° of water (Marinus of Tyre, 70 AD).Marinus of Tyre Columbus believed that 1° represented only 56 miles (Alfraganus, XIth century)Alfraganus He knew there was an island named Japan off the cost of China… Model Circumference of the Earth as 25,255 km at most, Canary Island to Japan : 3,700 km (Reality: 12,000 km.)

122. The More Structures The Merrier Average Improvement over T-Coffee Struc/Seq Ratio

123. The Right Mixt of Methods

124. 3D-Coffee: Combining Sequences and Structures Within Multiple Sequence Alignments

125. Applications

126. Looking-Up The DNA Behind The Sequences: PROTOGENE

127. SAR Analysis Correlate Alignment Variations with Reactivity Application to the Human Kinome Collaboration with Sanofi-Aventis Main Issue: – Training problem  Proper Benchmarking

128. ncRNA Multiple Alignments with R-Coffee Laundering the Genome Dark Matter Cédric Notredame Comparative Bioinformatics Group Bioinformatics and Genomics Program

129. No Plane Today…

130. ncRNAs Comparison And ENCODE said… “nearly the entire genome may be represented in primary transcripts that extensively overlap and include many non-protein-coding regions” Who Are They? – tRNA, rRNA, snoRNAs, – microRNAs, siRNAs – piRNAs – long ncRNAs (Xist, Evf, Air, CTN, PINK…) How Many of them – Open question – 30.000 is a common guess – Harder to detect than proteins.

131. ncRNAs can have different sequences and Similar Structures

132. ncRNAs are Difficult to Align Same Structure  Low Sequence Identity Small Alphabet, Short Sequences  Alignments often Non- Significant

133. Obtaining the Structure of a ncRNA is difficult Hard to Align The Sequences Without the Structure Hard to Predict the Structures Without an Alignment

134. The Holy Grail of RNA Comparison: Sankoff’ Algorithm

135. The Holy Grail of RNA Comparison Sankoff’ Algorithm Simultaneous Folding and Alignment – Time Complexity: O(L 2n ) – Space Complexity: O(L 3n ) In Practice, for Two Sequences: – 50 nucleotides: 1 min.6 M. – 100 nucleotides 16 min.256 M. – 200 nucleotides 4 hours 4 G. – 400 nucleotides3 days3 T. Forget about – Multiple sequence alignments – Database searches

136. The next best Thing: Consan Consan = Sankoff + a few constraints Use of Stochastic Context Free Grammars – Tree-shaped HMMs – Made sparse with constraints The constraints are derived from the most confident positions of the alignment Equivalent of Banded DP

137. Going Multiple…. Structural Aligners

138. Game Rules Using Structural Predictions – Produces better alignments – Is Computationally expensive Use as much structural information as possible while doing as little computation as possible…

139. Adapting T-Coffee To RNA Alignments

140. T-Coffee and Concistency…

144. Consistency: Conflicts and Information X Y X Z Y WZ X Z Y Z W Y W X Z Y Z X W Y Z X W Partly Consistent  Less Reliable Fully Consistent  More Reliable Y is unhappy X is unhappy X Y

145. RNA Sequences Secondary Structures Primary Library R-Coffee Extended Primary Library Progressive Alignment Using The R-Score RNAplfold Consan or Mafft / Muscle / ProbCons R-Coffee Extension R-Score

146. C C R-Coffee Scoring Scheme G G R-Score (CC)=MAX(TC-Score(CC), TC-Score (GG))

147. Validating R-Coffee

148. RNA Alignments are harder to validate than Protein Alignments Protein Alignments  Use of Structure based Reference Alignments RNA Alignments  No Real structure based reference alignments – The structures are mostly predicted from sequences – Circularity

149. BraliBase and the BraliScore Database of Reference Alignments 388 multiple sequence alignments. Evenly distributed between 35 and 95 percent average sequence identity Contain 5 sequences selected from the RNA family database Rfam The reference alignment is based on a SCFG model based on the full Rfam seed dataset (~100 sequences).

150. BraliBase SPS Score RFam MSA Number of Identically Aligned Pairs SPS= Number of Aligned Pairs

151. BraliBase: SCI Score RNApfoldRNApfold (((…)))…((..))  G Seq1 (((…)))…((..))  G Seq2 (((…)))…((..))  G Seq 3 (((…)))…((..))  G Seq4 (((…)))…((..))  G Seq5 (((…)))…((..))  G Seq6 RNAlifold (((…)))…((..)) ALN  G Average  G Seq X Cov  G ALN SCI= Covariance

152. BRaliScore Braliscore= SCI*SPS

153. RM-Coffee + Regular Aligners MethodAvg BraliscoreNet Improv. direct +T+R +T+R ----------------------------------------------------------- Poa0.620.650.70 48154 Pcma0.620.640.67 34120 Prrn0.640.610.66-63 45 ClustalW0.650.650.69 -7 83 Mafft_fftnts0.680.680.72 17 68 ProbConsRNA0.690.670.71-49 39 Muscle0.690.690.73-17 42 Mafft_ginsi0.700.680.72-49 39 ----------------------------------------------------------- RM-Coffee40.71 /0.74 / 84

154. How Best is the Best…. M-Locarna234 ***183 ** Stral169 *** 62 FoldalignM146 61 Murlet130 *-12 Rnasampler129 *-27 T-Lara125 *-30 Poa241 ***217 *** T-Coffee241 ***199 *** Prrn232 ***198 *** Pcma218 ***151 *** Proalign216 ***150 ** Mafft fftns206 ***148 * ClustalW203 ***136 *** Probcons192 ***128 * Mafft ginsi170 ***115 Muscle169 ***111 Method vs. R-Coffee-Consan vs. RM-Coffee4

155. Range of Performances Effect of Compensated Mutations

156. Conclusion/Future Directions T-Coffee/Consan is currently the best MSA protocol for ncRNAs Testing how important is the accuracy of the secondary structure prediction Going deeper into Sankoff’s territory: predicting and aligning simultaneously

157. www.tcoffee.org cedric.notredame@europe.com Credits and Web Servers Andreas Wilm Des Higgins Sebastien Moretti Ioannis Xenarios Cedric Notredame CGR, SIB, UCD

158. Prank Vs Prank

159. Gop=0 Gep=0

160. Prank Vs Prank The reconstruction of evolutionary homology -- including the correct placement of insertion and deletion events -- is only feasible for rather closely-related sequences. PRANK is not meant for the alignment of very diverged protein sequences. If sequences are very different, the correct homology cannot be reconstructed with confidence and http://www.ebi.ac.uk/goldman -srv/prank/

161. Do Benchmarks All Tell the same story? Based on

162. Probcons: Different Primary Library Score=  (MIN(xz,zk))/MAX(xz,zk) Score(xi ~ yj | x, y, z)  ∑k P(xi ~ zk | x, z) P(zk ~ yj | z, y)