7: Structure-Sequence Relationship

Question 1

How does protein sequence affect protein structure and function?

Answer 1

Protein sequence determines protein structure, which is linked to its protein function.

Question 2

What are the key concepts in sequence-structure relation?
1+3

Answer 2

Protein sequence determines structure
Unique fold: single lowest energy state
Stable fold: resistant to small environmental changes
Accessible fold: can reach stable fold normally

Question 3

Unique fold tests (3)

Answer 3

denatured and refolded into same fold
different conditions result in same fold.
> not dependant on environment
structure determination w/ various methods w/ various conditions results in same fold.

not dependant on environment
structure determination w/ various methods w/ various conditions results in same fold.

Question 4

Unique fold limitations (3)

Answer 4

Multi domain proteins hard to refold in vitro.
Beta amyloids (proteins unfold and aggregate to these instead).
1% of proteins switch fold upon stimuli.

Question 5

What is the energy landscape of protein folding like?

Answer 5

It resembles a rugged-surface funnel with shallow local minima.

Question 6

For accessible folds is energy consuming assistance required?

Answer 6

No, they should not be required to reach the lowest energy state.

but chaperones can increase the folding efficiency (with and without energy consumption) to avoid local minima.

Question 7

What is the correlation between sequence identity and protein folds?

Answer 7

Proteins with >20% sequence identity share folds.

provided from stability against mutations
exceptions can be made tho where 1 single AA change can result in fold change.

Question 8

Protein oligomerisation (4), multi domain structures, enzyme specificity are less correlated to sequence than principle domain folds?

Answer 8

True.
oligomers and multi domain interfaces typically only involve a small amount of all residues
= less dependant on sequence

and enzyme specificity can be determined by single AA
= mutations are bad

only at the domain level is the fold linked to its sequence.

Question 9

What factors are less correlated to sequence than domain folds?

Answer 9

Protein oligomerization and enzyme specificity.

Question 10

What physicochemical parameters can be predicted from single sequences?

Answer 10

Isoelectric Point (pI) and extinction coefficient.

weighted sum of all AA.

Question 11

What other predictions can be made from single sequences?

Answer 11

Linear motifs.*

Secondary structure.
based on tendency of AA to occur in specific secondary structures.
60% accurate.

short sequence patterns within proteins that act as functional signals e.g. kinase recognition site.

Question 12

Sequence known vs Structures known

Answer 12

1000x more sequences known than structures.

using seq. data and evolutionary relationships is important for seq-struc relationship analysis.

Question 13

Sequence similarity vs sequence identity w/ example

Answer 13

Sequence similarity measured often by physiochemical property.

Sequence identity w/ AA seq.

e.g. 30% similar with respect to AA hydrophobicity.

60% identical.

Question 14

What are the types of gene relationships in evolution?

Answer 14

Homolog genes: from common ancestor
Ortholog genes: separated by speciation
Paralog genes: from duplication events within species
Analog genes: similar functions, different origins

Question 15

Evolutionary events:

Answer 15

InDel,
Mutation.
= missense(=aa change), nonsense (=stop), silent (=no change).
Rearrangements: incl indel, inversions, duplication etc.

Question 16

What is the purpose of sequence alignments?

Answer 16

To match sequences by adjusting for evolutionary events.
> introducing indels.
optimal arrangement defined by residue-pair scoring and overall scoring.

Question 17

Identity scoring matrix pro/con

Answer 17

+ easy to compute.
- doesnt account for aa similarity

Question 18

Physiochemical scoring matrix pro/con

Answer 18

+ good for struc/func analysis.
- inferior to other methods

Question 19

What is the Needleman-Wunsch algorithm used for?

Answer 19

Calculating pairwise GLOBAL sequence alignment.
incl gap penalties.

Question 20

What is the Smith-Waterman algorithm used for?

Answer 20

Calculating pairwise LOCAL sequence alignment.
start from highest value and work back.
no gap penalty.

Question 21

Pairwise sequence alignments, pros cons

Answer 21

+ mathematically defined optimal solution
- not feasible for multiple sequences (computing time inc exponentially)

Question 22

What is the purpose of heuristic methods in sequence alignment?

Answer 22

To perform faster multiple sequence alignments

Question 23

What heuristic method is established for faster multiple sequence alignments?

Answer 23

CLUSTAL

Calculate pairwise alignments
Build a phylogenetic tree
Perform progressive alignment based on the tree

Question 24

How are large sets of sequences represented for progressive alignments/CLUSTAL?

Answer 24

As Profiles
As linear Hidden Markov Models (HMMs)

Question 25

What does a Markov chain describe? What is a characteristic of a linear Hidden Markov Model (HMM)?

Answer 25

A sequence of possible events with dependencies LHMM: Walks through the model in one direction

Question 26

What advantages do profiles and HMMs provide?

Answer 26

Represent information of sets of aligned sequences Improve predictions based on multiple alignments

Question 27

What are the benefits of using HMM alignments combined with known structural data?

Answer 27

extracting features from large sets of proteins >> single seq + combined with known data = Improved secondary structure prediction Prediction of disordered regions Prediction of domain boundaries Prediction of membrane protein topology

Question 28

What are the steps in homology modeling?

Answer 28

Identify homologous template structures Align target sequence based on template structure Re-model insertions and deletions Refine and validate model > used to predict entire structure

Question 29

At what percentage of sequence identity is homology modeling typically trustworthy?

Answer 29

Greater than 20% = high quality models > non homologous regions = harder to model e.g. loops/side chains

Question 30

What is threading and What is the risk associated with threading for fold identification?

Answer 30

threading = identification without >20% identity templates available results in Globally wrong fold identification

Question 31

what is and What does co-evolutionary analysis reveal about amino acids?

Answer 31

Correlation of mutations in pairs of amino acids = if X->B how likely is Y->C Direct information on protein fold and secondary structure

Question 32

why not do more than pairs for co-evolutionary analysis?

Answer 32

higher order correlations contribute little in addition because 2nd order correlation between pairs is sufficient

Question 33

Why is there correlation between pairs of amino acids?

Answer 33

often due to direct contact in 3D structure between amino acids that are far in sequence

Question 34

What is required for statistical significance in co-evolutionary analysis?

Answer 34

A large number of aligned sequences

Question 35

What does the output of Co-evolutionary analysis provide for modeling 3D structure?

Answer 35

Potential residue contacts + Visualization of secondary structural elements

Question 36

What is the role of Deep Mind AlphaFold in protein structure prediction? Output?

Answer 36

Predicts protein structure using deep learning Trained on structures from the PDB Output: Inter-residue distances and torsional angles

Question 37

How does AlphaFold 2 differ from its predecessor? output?

Answer 37

Uses a different neural network architecture Final output: 3D structure w/ iterative refinement

Question 38

Alphafold multimer

Answer 38

predicting oligomeric complexes (2+ proteins) has high error rate

Question 39

What are the properties of predictions made by AlphaFold? not super important

Answer 39

Accuracy decreases with fewer than 30 sequences Improvements in MSA depth lead to small gains pairwise residue distance accuracy predicted handles missing physical context trained to produce protein struc most likely to appear as paer of a PDB structure using whole sets is computationally challenged

Question 40

What are the principles, accuracy, and limitations of homology modeling and threading?

Answer 40

Homology modeling: Trustworthy at >20% identity Threading: Higher risk of incorrect fold identification

Question 41

What are the requirements and advantages of co-evolutionary analysis?

Answer 41

Requires large aligned sequences Provides insights into protein structure and interactions