Structural bioinformatics

Question 1

Why do we need protein-protein docking?

Answer 1

PDB contains ~4,000 hetero-dimers (including duplicate copies for same complex)
 Compare to number of entries ~170,000
 Can one predict the structure of a complex starting with
unbound components?
 Unbound components can be experimental or high quality predicted structures
 Need to be able to model limited conformational change.

Question 2

Describe typical protein protein docking

Answer 2

Question 3

Walk through protein docking step by step

Answer 3

1. Global search to find goo,d overlap of surfaces of protein
2. Residue residue interactions - score with empirical residue park potential. You need to check if a given pair is present in the protein more frequently than just by chance
3. Search for clusters of similar complex geometry with low energy. Many more ways to be reign than correct so correct solution will be found much more often than any individual wrong solution.
4. Refinement - search for optimal combination of side-chain rotamers by energy calculation.
5. Functional residue information - the function can give you info about the structure

Question 4

Template based prediction from homologous structure of a complex

Answer 4

X-ray structure of protein A’ complexed with protein B’
A’ is homologue of A B’ is homologue of B
If A/B interface is favourable evaluated in 3D then predict A interacts with B

Question 5

Template based modelling - sequence search

Answer 5

Start with sequence protein A and protein B
 Based on sequence similarity, search library of complexes in PDB for a complex A’ / B’ where A is homologous to A and B is homologous to B’
 Align sequence A to A’ and B to B’
 Sequence search via BLAST, PSIBLAST or an advanced statistical model known as Hidden Markov Model HMM

Question 6

Template based modelling - 2 model construction

Answer 6

On 3D structure of complex change sequence from A’ to A and B’ to B
 Adjust any loops where there is an insertion or deletion  Refine complex

Question 7

Steps template-based modelling 3 – alternate model selection

Answer 7

Sometimes there can be several suitable templates as several have similar sequence identity
 Construct several models
 Score models (similar to ab initio docking)
 Choose best model
 NB this is one approach but several variations in template-based modelling

Question 8

Coevolution and protein interactions

Answer 8

 Concept of correlated mutations extend to homo and hetero complexes
 AlphaFold multimer and Colab can consider complexes  Active area of research – results very encouraging

Question 9

What is the gene ontology?

Answer 9

A controlled vocabulary that can be applied to all organisms
 Used to describe gene products - proteins and RNA - in any organism
 All descriptions are supported by some level of evidence

Question 10

How does GO work?

Answer 10

It captures information about 3 important features of function:
 What does the gene product do?
 Why does it perform these activities?  Where does it act?

Question 11

Describe the 3 gene ontologies

Answer 11

 Molecular Function = biochemical function
 the tasks performed by individual gene products; examples
are carbohydrate binding and GTPase activity
 Biological Process = biological goal or objective (higher level function)
 broad biological goals, such as mitosis or purine metabolism, that are accomplished by combinations of individual molecular functions.
 Cellular Component = active location
 subcellular structures, locations, and macromolecular complexes; examples include nucleus, telomere, and RNA polymerase II holoenzyme

Question 12

Can you compare gene ontologies?

Answer 12

No, each branch is independent so you can’t really compare

Question 13

Where do annotations come from?

Answer 13

Annotation source is important
 It enables you to assess how confident the annotation is
 GO associates annotations with an evidence code that indicates its source.

Question 14

Uses of GO

Answer 14

 Enhanced predictors of protein function return prediction of GO terms
 Common features in a set of over-expressed / under- expressed genes can be reported as belonging to a common GO group

Question 15

Why do you want to computionally predict protein function

Answer 15

About 240M sequences but time taken experimentally to determine function can be several years

Question 16

Approches to function prediction

Answer 16

General homology search
 BLAST (single sequence information)
 PSIBLAST (multiple sequence information in profile)
 HMMs scan (multiple sequence information in hidden Markov model)

Question 17

Accuracy of database searching

Answer 17

Question 18

Orthologues and paralouges and their EC numbers

Answer 18

Orthologues: same EC and high %id
Paralogues: Different EC and lower % id

Question 19

Relationship %id and functional transfer

Answer 19

To some extent these concepts about EC numbers apply to all types of function but no clear cut rules
 Function challenging to quantify
 Start with a query sequence
 Perform BLASTP search

Question 20

What can you predict based on the %id between two proteins?

Answer 20

 If find a protein from another species and >~50% identity could easily be an orthologue with same function
 If find a protein from another species and >~30% identity could be a paralogue and have related function
 If find a protein from same species and >30%id could be a paralogue and have related functions

Question 21

Further complications - Domains

Answer 21

• Confident match by a search program
• Search programs identify local similarities. Two sequences may share a region/domain of similarity but also other domains that are different.
• But not all domains are shared – function probably different

Question 22

NetGo – Using different sources (Details not important)

Answer 22

Question 23

Structural searching

Answer 23

 Use structural searching to identify similar fold (e.g. DALI ot in CATH))
 Similar 3D structure may suggest related function
 If functional site in match is known examine if similar residues in your newly-determined protein

Question 24

Can finding a similar fold tell you anything about function?

Answer 24

No, to find about function you need to dig a little deeper

Question 25

Convergence of serine protease active site

Answer 25

Trypsin and subtilisin have quite different 3D folds
 Trypsin and subtilisin have totally different sequences
 Same three active site residues (Ser, His, Asp) located in very similar positions
 Order along chain of Ser, His, Asp different
 Infer convergence of function
 Favourable arrangement of residues to perform a function that arose vias evolution independently

Question 26

Principles for non polar proteins

Answer 26

 Apart from the ends, the protein chain in the membrane is in a non-polar (i.e. hydrophobic) environment.
 Side-chains tend to be hydrophobic
 Main-chain cannot have NH and CO groups not forming
hydrogen bonds
 Single spanning membrane regions one α-helix
 Most (but not all) membrane bound proteins formed from α-helical segments

Question 27

Transmembrane helices

Answer 27

 Hydrophobic section of membrane about 35 – 50A wide
 Section not translocated (i.e. pass through membrane) that
abuts membrane often contains + ve charges
 Each residue in an a-helix advances structure by 1.8A
 Transmembrane helices tend to be between 20-30 residues long

Question 28

Simple methods to identify transmembrane regions

Answer 28

 Early methods searched for runs of very non-polar (hydrophobic) residues along squence
 Use a scale of how hydrophobic each residue is  Hopp & Woods scale
 Kyte & Doolitle scale
 Calculate average over a window of several residues (typically 11)

Question 29

Signal peptides

Answer 29

Signal peptides refer to the sequence at start of proteins ranging from 15 – 60 residues that directs the protein to the correct cellular location.
 Generally signal peptides cleaved off
 Often has a hydrophobic region followed by a pattern typical of
the cleavage site
 In predictions need to distinguish transmembrane regions from signal peptides

Question 30

Current methods for transmembrane predictions

Answer 30

 Until recently prediction methods developed HMMs (hidden Markov models) based on aligned sequences
 DeepTMHHM – The algorithm predicts how the sequence maps onto
the different potential from the N- to the C- terminus a deep learning approach that predicts transmembrane structures and signal peptides,

Question 31

Low complexity regions

Answer 31

 Regions with composition biased strongly to a small number of amino acids
 e.g. RRRRRRR, RSSRSRSSRSR, GGSSSSDDS
 Occur in the sequences of a significant number of
proteins
 Distort statistical significance scores of alignments
 The program SEG (by Wootton) is often used
 Replaces low complexity regions with lower case in BLAST searches at NCBI

Question 32

Coiled-coils

Answer 32

Two or three intertwined α-helices
 Can be short segment (20 residues) or far longer  Identified using COILS written by Lupas

Question 33

Disordered proteins

Answer 33

 Some globular proteins have small regions which are disordered and cannot be identified by crystallography or NMR – often N- or C- terminus or a long loop.
 Some proteins do not adopt a single structure but are highly flexible
 Often protein becomes structured when protein binds to another protein

Question 34

Prediction of disordered regions

Answer 34

 Based on principles– tend to have lower fraction of hydrophobic residues than folded protein with a hydrophobic core.
 Machine learning based- Most programs now based on neural networks, support vector machines
 PONDR-FIT (Dunker)
 DISOPRED2 (Jones, UCL)
 IUPRED
 (see Expasy page – list of programs)
 Use AlphaFold prediction where confidence given by pLDDT<70% (see Piovesan, et al Protein Science, 31(11), e4466.)