Applied Bioinformatics

Week 11

Topics

• Protein Secondary Structure

• RNA Secondary Structure

Theory I

Recall Domains

• Functional region of a protein sequence

• Proteins may have several domains

• Generally identified by MSA

Domains

• Convey function

• Function derives from 3D structure

• How to determine 3D structure of proteins?

• First step secondary structure

Four levels of protein structure

Structure

Secondary Structure

• Local three dimensional structure

• Elements– Helix– Sheet– Coil

G = 3-turn helix (310 helix). Min length 3 residues.H = 4-turn helix (α helix). Min length 4 residues.I = 5-turn helix (π helix). Min length 5 residues.T = hydrogen bonded turn (3, 4 or 5 turn)E = extended strand in parallel and/or anti-parallel  β-sheet conformation. Min length 2 residues.B = residue in isolated β-bridge (single pair β-sheet hydrogen bond formation)S = bend (the only non-hydrogen-bond based assignment)

Secondary Structure 8 different categories

(DSSP):H: - helixG: 310 – helixI: - helix (extremely

rare) E: - strandB: - bridgeT: - turnS: bend L: the rest

Protein Secondary Structure [3]

Alpha Helix-

Structure repeats itself evry5.4 Angstroms along the helix axis

Every main chain CO and NH group is hydrogen bonded to a peptide bond 4 residues away

Beta Sheet – Two or more polypeptide chains run alongside each other and are linked by hydrogen bonds

Yuchun Tang, Preeti Singh, Yanqing Zhang, Chung-Dar Lu and Irene Weber, Georgia State University

Simplification

• 20 amino acids

• 5 - 11 groups of amino acids– Amino acids with similar chemical properties– Depends on the study

• 3 secondary structures

Secondary Structure Preditiction

• Sheet/ helix forming tendency of amino acids– Up to 60% accurate

• MSA -> neighborhood exploitation– Words of several aa are formed– Hydrophobicity is included– Up to 80% accurate

Propensities

Generation of Prediction Methods

• 1st generation : single residue statistics – Base on single amino acid propensity

• 2nd generation : segment statistics – Propensity for segments of 3-51 adjacent residues

• 3rd generation : evolution to better predictions – The use of evolutionary information (evolutionary

profile)

Assignment to Structure

• Sliding window of 7 amino acids– Why 7?

• Middle amino acid is assigned average propensity– Helix, Sheet

• Long stretches of similar assignments

About 2 turns (3.6 per turn)

Example: Window • Consider a secondary structure (x, e) and the window of

length 5 with the special position in the middle (bold letters)

• Fist position of the window is:

x = A R N S T V V S T A A . . .

e = ? ? H H C C C E E E . . . .

Window returns instance:

A R N S T H

Example: Window • Second position of the window is:

x = A R N S T V V S T A A . . .

e = ? ? H H C C C E E E . . . .

• Windows returns instance: R N S T V H

• Next instances are:N S T V V C

S T V V S C

T V V S T C

Practical Secondary Structure Prediction

• Can aid in MSA– If structures are not more similar than the

aligned sequences; there is a problem

• Step towards three dimensional structure

• Clue about architecture– 28 regular protein architectures

PSIPRED Example

Secondary structure prediction methods

PSI-pred (PSI-BLAST profiles used for prediction; David Jones, Warwick)

JPRED Consensus prediction (includes many of the methods given below; Cuff & Barton, EBI)

DSC King & SternbergPREDATORFrischman & Argos (EMBL) PHD home page Rost & Sander, EMBL, Germany ZPRED server Zvelebil et al., Ludwig, U.K. nnPredict Cohen et al., UCSF, USA. BMERC PSA Server Boston University, USA SSP (Nearest-neighbor) Solovyev and Salamov, Baylor College, USA.

http://speedy.embl-heidelberg.de/gtsp/secstrucpred.html

Andrew CR Martin, UCL

Consensus prediction method

hydrophobichighly conservedb= buried, e = exposed

Andrew CR Martin, UCL

Consensus prediction method -JPRED

hydrophobichighly conservedb= buried, e = exposed

amphipathic

hydrophobic

Andrew CR Martin, UCL

Neural network prediction - PHD

Multiple alignment

of protein family

SS profile for window of adjacent residues

Andrew CR Martin, UCL

Hidden Markov Models-HMMSTR

amino acid

secondary structure element

structural context

Markov state

• Recurrent local features of protein sequences

• Accuracy of 74%

Bystroff et al., 2000Andrew CR Martin, UCL

Consensus/ Meta Prediction Method

• Uses more than one existing method

• Learns how to combine the results

• Produces a result which is on average better than the single methods

• E.g.: http://gor.bb.iastate.edu/cdm/

Prediction Accuracy Assessment

• Protein Structure Prediction Center – http://predictioncenter.org/

• CASP– Critical Assessment of protein Structure

Prediction

Hydrophobicity

Assignment to Structure

• Sliding window of 5-7 or 19-21 amino acids– Why?

• Otherwise same idea as for secondary structure forming propensities

End Theory I

Mindmapping

10 min break

Practice I

Sec Struct Predictionhttp://bioinf.cs.ucl.ac.uk/psipred/psiform.htmlhttp://compbio.soe.ucsc.edu/HMM-apps/T02-query.html http://distill.ucd.ie/porter/ http://sable.cchmc.org/ http://www.compbio.dundee.ac.uk/www-jpred/advanced.html http://genamics.com/expression/strucpred.htm http://www.predictprotein.org/ http://npsa-pbil.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_phd.html http://www.chemie.uni-erlangen.de/lanig/PMII/sek_str.html http://npsa-pbil.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_sopma.html http://molbiol-tools.ca/Protein_secondary_structure.htm http://mobyle.pasteur.fr/cgi-bin/portal.py?form=predator http://www.aber.ac.uk/~phiwww/prof/ http://www.expasy.ch/tools/ http://gor.bb.iastate.edu/ http://www.predictprotein.org/

In class assignment• Choose a protein sequence

– Not too short!• Perform secondary structure predictions with as

many tools as possible– Google at least one more than given in the slides

• Retrieve and rewrite the predictions such that they use the 3 letter code (H,C,S; Helix, Coil, Sheet)– Use search and replace functionality of your word

processor• Make an MSA with the predicted secondary

structures to compare the results– Are there gaps? – Are they within the transition from one secondary

structure to the next?

Try to predict TMDs

• Find a protein with TMDs

• Expasy will provide you with prediction methods– DAS - Prediction of transmembrane regions in prokaryotes using the Dense

Alignment Surface method (Stockholm University)– HMMTOP - Prediction of transmembrane helices and topology of proteins

(Hungarian Academy of Sciences)– PredictProtein - Prediction of transmembrane helix location and topology

(Columbia University)– SOSUI - Prediction of transmembrane regions (Nagoya University, Japan)– TMHMM - Prediction of transmembrane helices in proteins (CBS; Denmark)– TMpred   - Prediction of transmembrane regions and protein orientation (EMBnet-

CH)– TopPred - Topology prediction of membrane proteins (France)

End Practice I

Theory II

RNA

• Coding RNA– Results in protein

• Non Coding RNA– Structural– Regulational– Catalytic– …

RNA Basicstransfer RNA (tRNA)

messenger RNA (mRNA)

ribosomal RNA (rRNA)

small interfering RNA (siRNA)

micro RNA (miRNA)

small nucleolar RNA (snoRNA)

http://www.genetics.wustl.edu/eddy/tRNAscan-SE/

RNA Secondary Structure

• Just like amino acids interact to form a secondary structure, nucleotides do the same

• Here base pairing is the driving motor

• Generally the structure of RNA molecules is projected onto 2 dimensions

Chemical Structure of RNAFour base types.

Distinguishable ends.

Partial Tertiary Structure

One illustration

Yet Another Tertiary Structure

Found via google

Our Final Tertiary Picture

Very complex

A Partial RNA Secondary Structure

Pure Secondary Structure

RNA Folding

• Single stranded RNA– Unstable– Base pairs with complementary

sequences– Base pair stacking– Favorable loop sizes

• Highest Stability– Lowest energy model

• Folding process– Not known in detail– Extremely fast

RNA Secondary Structure Prediction

Dynamic Programming Approaches

Sarah Aerni

http://www.tbi.univie.ac.at/

OutlineRNA folding

Dynamic programming for RNA secondary structure prediction

Covariance model for RNA structure prediction

RNA Secondary Structure

Hairpin loopJunction (Multiloop)

Bulge Loop

Single-Stranded

Interior Loop

Stem

Image– Wuchty

Pseudoknot

Sequence Alignment as a method to determine structure

Bases pair in order to form backbones and determine the secondary structure

Aligning bases based on their ability to pair with each other gives an algorithmic approach to determining the optimal structure

Base Pair Maximization – Dynamic Programming Algorithm

Simple Example:Maximizing Base Pairing

Base pair at i and jUnmatched at iUmatched at jBifurcation

Images – Sean Eddy

S(i,j) is the folding of the subsequence of the RNA strand from index i to index j which results in the highest number of base pairs

Base Pair Maximization – Dynamic Programming Algorithm

Alignment Method Align RNA strand to itself Score increases for feasible base

pairs

Each score independent of overall structure

Bifurcation adds extra dimension

Initialize first two diagonal arrays to 0

Fill in squares sweeping diagonally

Images – Sean Eddy

Bases cannot pair, similarto unmatched alignment

S(i, j – 1)

Bases can pair, similarto matched alignment

S(i + 1, j)

Dynamic Programming – possible paths S(i + 1, j – 1) +1

Base Pair Maximization – Dynamic Programming Algorithm

Alignment Method Align RNA strand to itself Score increases for feasible base

pairs

Each score independent of overall structure

Bifurcation adds extra dimension

Initialize first two diagonal arrays to 0

Fill in squares sweeping diagonally

Images – Sean Eddy

Reminder:For all k

S(i,k) + S(k + 1, j)

k = 0 : Bifurcation max in this case

S(i,k) + S(k + 1, j)

Reminder:For all k

S(i,k) + S(k + 1, j)

Bases cannot pair, similarBases can pair, similarto matched alignmentDynamic Programming –

possible pathsBifurcation – add values for

all k

Base Pair Maximization - Drawbacks

Base pair maximization will not necessarily lead to the most stable structureMay create structure with many interior loops or

hairpins which are energetically unfavorable

Comparable to aligning sequences with scattered matches – not biologically reasonable

Energy Minimization

Thermodynamic StabilityEstimated using experimental techniques

Theory : Most Stable is the Most likely

No Pseudknots due to algorithm limitations

Uses Dynamic Programming alignment technique

Attempts to maximize the score taking into account thermodynamics

MFOLD and ViennaRNA

Energy Minimization Results

Linear RNA strand folded back on itself to create secondary structure

Circularized representation uses this requirementArcs represent base pairing

Images – David Mount

All loops must have at least 3 bases in them Equivalent to having 3 base pairs between all arcs

Exception: Location where the beginning and end of RNA come together in circularized representation

Trouble with Pseudoknots

Pseudoknots cause a breakdown in the Dynamic Programming Algorithm.

In order to form a pseudoknot, checks must be made to ensure base is not already paired – this breaks down the recurrence relations

Images – David Mount

Energy Minimization Drawbacks

Compute only one optimal structure

Usual drawbacks of purely mathematical approachesSimilar difficulties in other algorithms

Protein structure

Exon finding

Alternative Algorithms - Covariaton

Incorporates Similarity-based methodEvolution maintains sequences that are importantChange in sequence coincides to maintain structure

through base pairs (Covariance)Cross-species structure conservation example – tRNA

Manual and automated approaches have been used to identify covarying base pairs

Models for structure based on resultsOrdered Tree ModelStochastic Context Free Grammar

Expect areas of basepairing in tRNA to be covarying betweenvarious species

Base pairing creates same stable tRNA structure in organisms

Mutation in one baseyields pairing impossible and breaksdown structure

Covariation ensuresability to base pair is maintained and RNAstructure is conserved

Binary Tree Representation of RNA Secondary Structure

Representation of RNA structure using Binary tree

Nodes represent

Base pair if two bases are shown

Loop if base and “gap” (dash) are shown

Pseudoknots still not represented

Tree does not permit varying sequences

Mismatches

Insertions & Deletions

Images – Eddy et al.

Covariance Model

HMM which permits flexible alignment to an RNA structure – emission and transition probabilities

Model trees based on finite number of states Match states – sequence conforms to the model:

MATP – State in which bases are paired in the model and sequence

MATL & MATR – State in which either right or left bulges in the sequence and the model

Deletion – State in which there is deletion in the sequence when compared to the model

Insertion – State in which there is an insertion relative to model

Transitions have probabilitiesVarying probability – Enter insertion, remain in current state, etc

Bifurcation – no probability, describes path

Covariance Model (CM) Training Algorithm

S(i,j) = Score at indices i and j in RNA when aligned to the Covariance Model

Independent frequency of seeing the symbols (A, C, G, T) in locations i or j depending on symbol.

Frequencies obtained by aligning model to “training data” – consists of sample sequences Reflect values which optimize alignment of sequences to model

Frequency of seeing the symbols (A, C, G, T) together in locations i and j depending on symbol.

Alignment to CM Algorithm

Calculate the probability score of aligning RNA to CM

Three dimensional matrix – O(n³)Align sequence to given subtrees in CM

For each subsequence calculate all possible states

Subtrees evolve from Bifurcations

For simplicity Left singlet is default

Images – Eddy et al.

•For each calculation take intoaccount the

• Transition (T) to next state • Emission probability (P) in the

state as determined by training data

Bifurcation – does not have a probabilityassociated with the stateDeletion – does not have an emission probability (P) associated with it

Images – Eddy et al.

Alignment to CM Algorithm

Covariance Model Drawbacks

Needs to be well trained

Not suitable for searches of large RNAStructural complexity of large RNA cannot be

modeled

Runtime

Memory requirements

End Theory II

Mindmapping

10 min break

Practice II

RNA Secondary Structure

• Online• http://compbio.cs.sfu.ca/taverna/alterna/• http://www.bioinfo.rpi.edu/applications/mfold/

• Download• RNAShapes• RNAFold

• Get RNAs– http://www.ncrna.org/frnadb/search.html