CSEP 527 Computational Biology. RNA: Function, Secondary Structure Prediction, Search, Discovery

Transcription

1 SEP 527 omputational Biology RN: Function, Secondary Structure Prediction, Search, Discovery

2 The Message ells make lots of RN noncoding RN Functionally important, functionally diverse Structurally complex New tools required alignment, discovery, search, scoring, etc. 2

3 Rough Outline Today Noncoding RN Examples RN structure prediction 3

4 RN DN: DeoxyriboNucleic cid RN: RiboNucleic cid Like DN, except: dds an OH on ribose (backbone sugar) racil () in place of thymine (T),, as before pairs with H 3 thymine uracil 4

5 RN Secondary Structure: RN makes helices too Base pairs 5 3 sually single stranded 5

6 Fig. 2. The arrows show the situation as it seemed in Solid arrows represent probable transfers, dotted arrows possible transfers. The absent arrows (compare Fig. 1) represent the impossible transfers postulated by the central dogma. They are the three possible arrows starting from protein. 6

7 Ribosomes Watson, ilman, Witkowski, & Zoller,

8 Ribosomes 1974 Nobel prize to Romanian biologist eorge Palade ( ) for discovery in mid 50 s proteins 3-4 RNs (half the mass) atalytic core is RN Of course, mrns and trns (messenger & transfer RNs) are critical too tomic structure of the 50S Subunit from Haloarcula marismortui. Proteins are shown in blue and the two RN strands in orange and yellow. The small patch of green in the center of the subunit is the active site. - Wikipedia 8

9 Transfer RN The adapter coupling mrn to protein synthesis. Discovered in the mid-1950s by Mahlon Hoagland ( , left), Mary Stephenson, and Paul Zamecnik ( ; Lasker award winner, right). 9

10 Bacteria Triumph of proteins 80% of genome is coding DN Functionally diverse receptors motors catalysts regulators (Monod & Jakob, Nobel prize 1965) 10

11 Proteins atalyze Biochemistry: Met Pathways 11

12 Proteins Regulate Biochemistry: The MET Repressor SM DN Protein lberts, et al, 3e. 12

13 lberts, et al, 3e. Not the only way! Protein way Riboswitch alternative SM rundy & Henkin, Mol. Microbiol 1998 Epshtein, et al., PNS 2003 Winkler et al., Nat. Struct. Biol

14 lberts, et al, 3e. Not the only way! Protein way Riboswitch alternatives SM-II SM-I orbino et al., enome Biol rundy, Epshtein, Winkler et al., 1998,

15 lberts, et al, 3e. Not the only way! Protein way Riboswitch alternatives SM-III SM-I SM-II Fuchs et al., NSMB 2006 rundy, Epshtein, Winkler et al., 1998, 2003 orbino et al., enome Biol

16 lberts, et al, 3e. Not the only way! Protein way Riboswitch alternatives SM-III SM-I SM-II SM-IV rundy, Epshtein, Winkler et al., 1998, 2003 orbino et al., enome Biol Fuchs et al., NSMB 2006 Weinberg et al., RN

17 lberts, et al, 3e. Not the only way! Protein way Riboswitch alternatives SM-III SM-I SM-II SM-IV rundy, Epshtein, Winkler et al., 1998, 2003 orbino et al., enome Biol Fuchs et al., NSMB 2006 Weinberg et al., RN 2008 Meyer, etal., BM enomics

18 nd in other bacteria, a riboswitch senses SH (SH) 18

19 ncrn Example: Riboswitches TR structure that directly senses/binds small molecules & regulates mrn widespread in prokaryotes some in eukaryotes & archaea, one in a phage ~ 20 ligands known; multiple nonhomologous solutions for some dozens to hundreds of instances of each on/off; transcription/translation; splicing; combinatorial control all found since ~2003; most via bioinformatics 19

20 20

21 New ntibiotic Targets? Old drugs, new understanding: TPP riboswitch ~ pyrithiamine lysine riboswitch ~ L-aminoethylcysteine, DL-4-oxalysine FMN riboswitch ~ roseoflavin Potential advantages - no (known) human riboswitches, but often multiple copies in bacteria, so potentially efficacious with few side effects? 21

22 ncrn Example: T-boxes 22

23 23

24 sed by Mfinder Found by scan hloroflexus aurantiacus eobacter metallireducens eobacter sulphurreducens hloroflexi δ -Proteobacteria Symbiobacterium thermophilum 24

25 ncrn Example: 6S medium size (175nt) structured highly expressed in E. coli in certain growth conditions sequenced in 1971; function unknown for 30 years 25

26 6S mimics an open promoter Bacillus/ lostridium ctinobacteria E.coli Barrick et al. RN 2005 Trotochaud et al. NSMB 2005 Willkomm et al. NR

27 Summary: RN in Bacteria Widespread, deeply conserved, structurally sophisticated, functionally diverse, biologically important uses for ncrn throughout prokaryotic world. Regulation of MNY genes involves RN In some species, we know identities of more riboregulators than protein regulators Dozens of classes & thousands of new examples in just the last 5-10 years 27

28 Vertebrate ncrns mrn, trn, rrn, of course PLS: snrn, spliceosome, snorn, teleomerase, microrn, RNi, SEIS, IRE, piwi-rn, XIST (X-inactivation), ribozymes, 28

29 MicroRN 1st discovered 1992 in. elegans 2nd discovered 2000, also. elegans and human, fly, everything between basically all multi-celled plants & animals nucleotides literally fell off ends of gels Hundreds now known in human may regulate 1/3-1/2 of all genes development, stem cells, cancer, infectious disease, 29

30 sirn 2006 Nobel Prize Fire & Mello Short Interfering RN lso discovered in. elegans Possibly an antiviral defense, shares machinery with mirn pathways llows artificial repression of most genes in most higher organisms Huge tool for biology & biotech 30

31 ncrn Example: Xist large ( 12kb) largely unstructured RN required for X-inactivation in mammals (Remember calico cats?) 31

32 Human Predictions Evofold S Pedersen, Bejerano, Siepel, K Rosenbloom, K Lindblad-Toh, ES Lander, J Kent, W Miller, D Haussler, "Identification and classification of conserved RN secondary structures in the human genome." PLoS omput. Biol., 2, #4 (2006) e33. 48,479 candidates (~70% FDR?) RNz S Washietl, IL Hofacker, M Lukasser, Hutenhofer, PF Stadler, "Mapping of conserved RN secondary structures predicts thousands of functional noncoding RNs in the human genome." Nat. Biotechnol., 23, #11 (2005) ,000 structured RN elements 1,000 conserved across all vertebrates. ~1/3 in introns of known genes, ~1/6 in TRs ~1/2 located far from any known gene FOLDLIN E Torarinsson, M Sawera, JH Havgaard, M Fredholm, J orodkin, "Thousands of corresponding human and mouse genomic regions unalignable in primary sequence contain common RN structure." enome Res., 16, #7 (2006) candidates from (of 100,000) pairs Mfinder Torarinsson, Yao, Wiklund, Bramsen, Hansen, Kjems, Tommerup, Ruzzo and orodkin. omparative genomics beyond sequence based alignments: RN structures in the ENODE regions. enome Research, Feb 2008, 18(2): PMID: candidates in ENODE alone (better FDR, but still high) 32

33 Bottom line? significant number of one-off examples Extremely wide-spread ncrn expression t a minimum, a vast evolutionary substrate New technology (e.g., RNseq) exposing more How do you recognize an interesting one? lue: onserved secondary structure 33

34 RN Secondary Structure: can be fixed while sequence evolves - 34

35 Why is RN hard to deal with? : Structure often more important than sequence 35

36 Structure Prediction

37 RN Structure Primary Structure: Sequence Secondary Structure: Pairing Tertiary Structure: 3D shape 37

38 RN Pairing Watson-rick Pairing - - ~ 3 kcal/mole ~ 2 kcal/mole ~1 kcal/mole Wobble Pair - Non-canonical Pairs (esp. if modified) 38

39 trn 3d Structure 39

40 trn - lt. Representations a.a. 3 5 nticodon loop nticodon loop 40

41 trn - lt. Representations nticodon loop nticodon loop 41

42 Definitions Sequence 5 r 1 r 2 r 3... r n 3 in {,,, T/} Secondary Structure is a set of pairs i j s.t. i < j-4, and no sharp turns if i j & i j are two different pairs with i i, then j < i, or i < i < j < j 2nd pair follows 1st, or is nested within it; no pseudoknots. 42

43 RN Secondary Structure: Examples base pair 4 ok sharp turn crossing 43

44 Nested Precedes Pseudoknot 44

45 pproaches to Structure Prediction Maximum Pairing + works on single sequences + simple - too inaccurate Minimum Energy + works on single sequences - ignores pseudoknots - only finds optimal fold Partition Function + finds all folds - ignores pseudoknots 45

46 Nussinov: Max Pairing B(i,j) = # pairs in optimal pairing of r i... r j B(i,j) = 0 for all i, j with i j-4; otherwise B(i,j) = max of: B(i,j-1) max { B(i,k-1)+1+B(k+1,j-1) i k < j-4 and r k -r j may pair} R Nussinov, B Jacobson, "Fast algorithm for predicting the secondary structure of single-stranded RN." PNS

47 Optimal pairing of r i... r j Two possibilities j npaired: Find best pairing of r i... r j-1 j i j-1 j Paired (with some k): Find best r i... r k-1 + best r k+1... r j-1 plus 1 Why is it slow? Why do pseudoknots matter? j i j-1 k-1 k k+1 47

48 Nussinov: omputation Order B(i,j) = # pairs in optimal pairing of r i... r j B(i,j) = 0 for all i, j with i j-4; otherwise B(i,j) = max of: B(i,j-1) max { B(i,k-1)+1+B(k+1,j-1) i k < j-4 and r k -r j may pair} K= Time: O(n 3 ) 48

49 Which Pairs? sual dynamic programming trace-back tells you which base pairs are in the optimal solution, not just how many 49

50 omputing one cell: OPT[2,18] =? n= 20 ( ( (.... ) ) ) ( ( (.... ) ) ) ase 1: ? no. ase 2: B 18 unpaired? lways a possibility; then OPT[2,18] 3 $ * 0 if i j 4 OPT(i, 0 0 j) 0 = 0 % 0 0 $ OPT[i, j -1] ' max 0 0 % & 1+ 0 max 0 0 t (OPT[i,t ] OPT[t , 0 0 j 1] 0 ( otherwise &* ) ((...))(...)...

51 omputing one cell: OPT[2,18] =? n= 20 ( ( (.... ) ) ) ( ( (.... ) ) ) $ 0 if i j * OPT(i, j) = % 0 0 $ OPT[i, j -1] ' max 0 0 % 1+ 0 max ( otherwise &* & t (OPT[i,t 1] + OPT[t +1, j 1] ) ase 3, 2 t <18-4: t = 2: no pair

52 omputing one cell: OPT[2,18] =? n= 20 ( ( (.... ) ) ) ( ( (.... ) ) ) $ 0 if i j * OPT(i, j) = % 0 0 $ OPT[i, j -1] ' max 0 0 % 1+ 0 max ( otherwise &* & t (OPT[i,t 1] + OPT[t +1, j 1] ) ase 3, 2 t <18-4: t = 3: no pair

53 omputing one cell: OPT[2,18] =? n= 20 ( ( (.... ) ) ) ( ( (.... ) ) ) $ 0 if i j * OPT(i, j) = % 0 0 $ OPT[i, j -1] ' max 0 0 % 1+ 0 max ( otherwise &* & t (OPT[i,t 1] + OPT[t +1, j 1] ) ase 3, 2 t <18-4: t = 4: yes pair OPT[2,18] (...(((...))))

54 omputing one cell: OPT[2,18] =? n= 20 ( ( (.... ) ) ) ( ( (.... ) ) ) ase 3, 2 t <18-4: t = 5: yes pair OPT[2,18] (..(((...)))) $ 0 if i j * OPT(i, j) = % 0 0 $ OPT(i, OPT[i, j -1) -1] ' max% ( otherwise &* & 1+ max t (OPT(i,t (OPT[i,t 1) 1] + OPT[t OPT(t +1, j 1] 1) )

55 omputing one cell: OPT[2,18] =? n= 20 ( ( (.... ) ) ) ( ( (.... ) ) ) ase 3, 2 t <18-4: t = 6: yes pair OPT[2,18] (.(((...)))) $ 0 if i j * OPT(i, j) = % 0 0 $ OPT(i, OPT[i, j -1) -1] ' max% ( otherwise &* & 1+ max t (OPT(i,t (OPT[i,t 1) 1] + OPT[t OPT(t +1, j 1] 1) )

56 omputing one cell: OPT[2,18] =? n= 20 ( ( (.... ) ) ) ( ( (.... ) ) ) ase 3, 2 t <18-4: t = 7: yes pair OPT[2,18] ((((...)))) $ 0 if i j * OPT(i, j) = % 0 0 $ OPT(i, OPT[i, j -1) -1] ' max% ( otherwise &* & 1+ max t (OPT(i,t (OPT[i,t 1) 1] + OPT[t OPT(t +1, j 1] 1) )

57 omputing one cell: OPT[2,18] =? n= 20 ( ( (.... ) ) ) ( ( (.... ) ) ) $ 0 if i j * OPT(i, j) = % 0 0 $ OPT[i, j -1] ' max 0 0 % 1+ 0 max ( otherwise &* & t (OPT[i,t 1] + OPT[t +1, j 1] ) ase 3, 2 t <18-4: t = 8: no pair

58 omputing one cell: OPT[2,18] =? n= 20 ( ( (.... ) ) ) ( ( (.... ) ) ) $ 0 if i j * OPT(i, j) = % 0 0 $ OPT[i, j -1] ' max 0 0 % 1+ 0 max ( otherwise &* & t (OPT[i,t 1] + OPT[t +1, j 1] ) ase 3, 2 t <18-4: t = 11: yes pair OPT[2,18] ((...))(...) (not shown: t=9,10, 12,13)

59 n= 20 ( ( (.... ) ) ) ( ( (.... ) ) ) omputing one cell: OPT[2,18] = 4 Overall, Max = 4 several ways, e.g.:..(...(((...)))) tree shows trace back: square = case 3 octagon = case 1 $ 0 if i j * OPT(i, j) = % 0 0 $ OPT[i, j -1] ' max 0 0 % 1+ 0 max ( otherwise &* & t (OPT[i,t 1] + OPT[t +1, j 1] )

60 pproaches to Structure Prediction Maximum Pairing + works on single sequences + simple - too inaccurate Minimum Energy + works on single sequences - ignores pseudoknots - only finds optimal fold Partition Function + finds all folds - ignores pseudoknots

61 Pair-based Energy Minimization E(i,j) = energy of pairs in optimal pairing of r i... r j E(i,j) = for all i, j with i j-4; otherwise E(i,j) = min of: E(i,j-1) energy of k-j pair min { E(i,k-1) + e(r k, r j ) + E(k+1,j-1) i k < j-4 } Time: O(n 3 )

62 Loop-based Energy Minimization Detailed experiments show it s more accurate to model based on loops, rather than just pairs Loop types Hairpin loop 2. Stack 3. Bulge 4 4. Interior loop 5. Multiloop 5

63 Zuker: Loop-based Energy, I W(i,j) = energy of optimal pairing of r i... r j V(i,j) = as above, but forcing pair i j W(i,j) = V(i,j) = for all i, j with i j-4 W(i,j) = min( W(i,j-1), min { W(i,k-1)+V(k,j) i k < j-4 } )

64 Zuker: Loop-based Energy, II hairpin stack bulge/ interior multiloop V(i,j) = min(eh(i,j), es(i,j)+v(i+1,j-1), VBI(i,j), VM(i,j)) VM(i,j) = min { W(i,k)+W(k+1,j) i < k < j } VBI(i,j) = min { ebi(i,j,i,j ) + V(i, j ) bulge/ interior i < i < j < j & i -i+j-j > 2 } Time: O(n 4 ) O(n 3 ) possible if ebi(.) is nice

65 Energy Parameters Q. Where do they come from? 1. Experiments with carefully selected synthetic RNs 2. Learned algorithmically from trusted alignments/structures [ndronescu et al., 2007]

66 Single Seq Prediction ccuracy Mfold, Vienna,... [Nussinov, Zuker, Hofacker, Mcaskill] Latest estimates suggest ~50-75% of base pairs predicted correctly in sequences of up to ~300nt Definitely useful, but obviously imperfect 50

67 pproaches to Structure Prediction Maximum Pairing + works on single sequences + simple - too inaccurate Minimum Energy + works on single sequences - ignores pseudoknots - only finds optimal fold Partition Function + finds all folds - ignores pseudoknots

68 pproaches, II omparative sequence analysis + handles all pairings (potentially incl. pseudoknots) - requires several (many?) aligned, appropriately diverged sequences Stochastic ontext-free rammars Roughly combines min energy & comparative, but no pseudoknots Physical experiments (x-ray crystallography, NMR) Next Lecture

69 Summary RN has important roles beyond mrn Many unexpected recent discoveries Structure is critical to function True of proteins, too, but they re easier to find from sequence alone due, e.g., to codon structure, which RNs lack RN secondary structure can be predicted (to useful accuracy) by dynamic programming Next: RN motifs (seq + 2-ary struct) wellcaptured by covariance models 51