Protein structure determination

http://www.thesgc.org/scientists/groups/oxford/crystallography

Preparation

• Protein purification
• Protein crystallization

Imaging

• X-ray crystallography
• Cryo EM! <– 3 to 5 Angstrom. Get projections, and reconstruct 3D structure that way!
• NMR protein structure determination. Get info about the distances between certain kinds of atoms, measured through the decay times of the correlations between nuclear spins.
• Small angle X-ray. rough shape, but very cheap!

Computational analysis

Uses Machine learning, Statistics

• Clustering, like UPGMA or K-means algorithm
• Statistical potential for analysis of Protein-decoy structures

http://www.ebi.ac.uk/interpro/

http://www.proteinmodelportal.org/

https://swissmodel.expasy.org/repository

Protein structure prediction

Protein alignment

Sequence alignment. Which amino acids match with which Structure alignment. Structure superposition. Minimize RMSD distance.

–> How to choose balance between more amino acids in the sequence alignment, and less RMSD. Balance depends on what we are more interested in, the geometry (bettet rmsd), or the biology (better alignment).

Structural alignment methods * rigid alignment * flexible alignment * elastic alignment. Change coordinate systems. Not yet good biological results.

All these methods are heuristic. They give different answers in general for complicated problems.

First modern protein comparison program. DALI. Representations: contact matrix.

–Meausre of how good alignment is

MAMMOTH. Representation: heptapeptides. Unit-vector rms. Likelihiox.

SCOP. Manual classification. A.Murzi * all alpha * all beta * alpha/beta. Mainly parallel beta sheets (beta-alpha-beta units) * alpha + beta. Mainly antiparallel beta sheets (segregated alpha and beta regions) * Folds: secondary structure aligns (ignoring loops) * Superfamily: evolutionary relation * family: CATH. Semi-manual classification FSSP. Automatic.

Fold space. PCA. Hou et al. Structure relatedness, evolution.