Protein Crystallography and Prediction - WIP Flashcards
What are the bonds in the primary, secondary and tertiary structures of a protein?
Primary structure: covalent, peptide, di-sulphide
Secondary structure: hydrogen, peptide, di-sulphide
Tertiary and quaternary structure: hydrophobic, hydrogen, ionic, van der Waals forces, di-sulphide
How can 3D structure of a protein be determined?
X-ray crystallography provides the clearest visualization (10nm-<0.1nm) and is the most accurate. Produces the clearest visualisation, 3D position of atoms. Shoot x-rays at crystal, detector collects data. Diamond light source is common, generates light using high energy electrons
Other methods:
NMR – second most accurate
Cryo Electron Microscopy
Computationally
What are the steps in protein crystallisation?
What is protein structure prediction?
Protein structure prediction is the inference of the 3D structure of a protein from its amino acid sequence – the prediction of secondary and tertiary structure from primary structure
Accurately predicted protein structures leads to:
- Antibody designs
- Employment for drug discovery
- Understanding protein-protein interactions
Describe Deep Learning Based protein structure prediction
A template free model, that relies on amino acid sequence and gene sequence data to predict the protein structure. Uses compilation of knowledge:
- Fragment reusability
- Protein structure formation
- Conserved regions
- Sequence alignments
- Co-evolving residues
Alpha fold-2 – computational method that can regularly predict protein structures with atomic accuracy
Describe Homology/Comparative modelling as protein structure prediction
Modelling of 3D structure of a protein by exploiting structural information from the known configurations of similar proteins. Compares target and template sequences.
The most accurate and successful approach to creative reliable protein structures when a suitable template exists, sufficient similarity is >30% (Structural conformation of a protein is more highly conserved than its amino acid sequence)
Phyre2 (Protein homology/analogy recognition engine) - uses HHblits (amino acid database) and protein structure database, and forms a hidden Markov model. Hidden Markov models exist for the known protein structure, and aligns with amino acid sequence, producing a final 3D structure model.
Describe threading and folding recognition for protein structure prediction
Predicts the structural fold of unknown protein sequences by fitting the sequence into a structural database by selecting the best fitting fold
Is carried out when there is not sufficient sequence similarity between the target and template to carry out comparative modelling
Not very common
What are the different methods of protein structure prediction?
Template free modelling (without a known structure protein) includes;
- Deep Learning based
- Physics based
- Secondary Structural Elements (alpha helix and beta sheet) based
- Fragment based
Based template modelling (with a protein with known structure to compare to) includes;
- Homology/Comparative modelling
- Threading and fold recognition
What is Alpha-fold?
A deep-learning based computational method that can regularly predict protein structures with atomic accuracy even in cases in which no similar structure is known. Almost a score of 90 in protein-folding contest, nearly equivalent to the experimentally determined structure
Alpha fold-2 works by predicting protein structure from amino acid sequence. The software generates multiple sequence alignments for the protein using a genetic and a structural database. Will produce a multiple sequence alignment for the protein, using pair presentation (how amino acids connect to each other). Evoformer network generates a structural model producing a 3D structure, and recycles this 3 times to increase accuracy of prediction
