Andrej Šali

Andrej Šali (born 1963, Kranj, Slovenia) is a computational structural biologist. He joined the faculty of the Rockefeller University in 1995, following his postdoctoral research at Harvard University. Since 2003, he has been the Professor and Vice Chair of the Department of Bioengineering and Therapeutic Sciences as well as the Scientific Director of California Institute for Quantitative Biosciences at University of California, San Francisco. He also serves as an editor for the journal Structure.

Education

Šali received his B.Sc. in Chemistry from the University of Ljubljana, 1987; his Ph.D. in Molecular Biophysics from Birkbeck College, University of London, 1991 (working with Tom Blundell); and did postdoctoral work at Harvard University (working with Martin Karplus).

Work

He is using computation grounded in the laws of physics and evolution to study the structure and function of proteins. For example, he developed comparative protein structure modeling by satisfaction of spatial restraints, implemented in the program MODELLER and integrative structure determination of macromolecular assemblies, implemented in the program IMP.

In July 2006, the magazine Business 2.0 ranked Šali and Stephen Maurer together at 41st place among "50 Who Matter Now".

Personal life

Šali is married to soprano (and former scientist) Heidi Moss and they have two daughters, Ava and Hana.

Citation Analysis

As at 11 August 2011, Šali has a total of 279 publications,^[1] 243 of which are found on the Web of Science.^[2] The average citation per item is 80.60 and the number of times cited is 19,585. Also, the average citation per year is 851.52. Šali has a Hirsch-index of 65, meaning 65 published items having at least 65 citations.

The Top Five Research Articles by Citation

How does a protein fold?^[3]

In this article, proteins were shown to fold from a denatured state to its native conformation in three steps. This process starts off with a rapid collapse of a denatured state (out of 1016 possible states) randomly into one of 1010 possible semi-compact molten globules. Thereafter, a slow search for a transition state (out of 103 possible states) occurs. The protein consequently adopts its stable native fold in a fast final step.

Comparative Protein Structure Modeling of Genes and Genomes^[4]

This is a review article on the steps in comparative protein structure modeling and how to perform evaluation to pick out errors in comparative models. The applications of such an endeavour are introduced and extension to whole genome (proteome) modeling is also discussed.

Protein Structure Prediction and Structural Genomics^[5]

This review article is somewhat similar to Comparative Protein Structure Modeling of Genes and Genomes in that it also discusses about comparative modeling, with the inclusion of de novo modeling as an alternative for target proteins without a suitable template. An example of a program that does de novo (or ab initio) protein structure prediction is Rosetta,^[6] which was written by Baker, the co-author of this article.

Modeling of Loops in Protein Structures^[7]

A shortcoming of comparative modeling is that homologous protein sequences often have variable loop regions which cannot be accurately modeled. Thus sampling and optimization are used to model loop conformations.

Most Important Algorithmic Contribution

Šali’s most important algorithmic contribution is the satisfaction of spatial restraints,^[8] which is used in his comparative protein structure modeling program called MODELLER. The algorithm is outlined in Figure 1.^[9]

The objective of this algorithm is to generate many restraints on the structure of the target sequence, using its alignment to template structures as a guide. This is based on the assumption that corresponding distances and angles between two pairs of aligned residues in the target and template structures are similar. These homology-derived restraints are usually supplemented by stereochemical restraints on bond lengths, bond angles, dihedral angles, and nonbonded atom-atom contacts obtained from a molecular mechanics force field. This is known as analytical supplementation. Restraints derived from experimental data, such as those found in a structure database of protein families, can act as empirical supplementation. The extraction of spatial restraints from such a database is shown in Figure 2.^[10]

Each spatial restraint can be expressed as a conditional probability density function (pdf)

obtained by smoothing a histogram of restraint frequencies. In Figure 1 for example, given that residues Q...T in the template align with residues W...D in the target, the Cα-Cα homology-derived restraint can be supplemented with Cα-Cα distance data of W and D from the database on the condition that they align with Q and T respectively [p((Cα(W)-Cα(D))/(Cα(Q)-Cα(T)...))]. Consequently, a frequency histogram of Cα(W)-Cα(D) restraints is obtained and can be used for satifaction of the corresponding homology-derived spatial restraints.

In the final step of the algorithm, optimization is used to minimize violations of all the spatial restraints. This starts off by constructing a model using the distance and angle restraints on the target sequence derived from its alignment with template 3-Dimensional (3D) structures. Then, local restraints for residues close in proximity are satisfied by the method of conjugate gradients, introducing more longer-ranged restraints iteratively. The resulting model is refined with Molecular Dynamics (MD) using simulated annealing. This whole process is shown in Figure 3.^[11]

Research Impact

MODELLER has enabled proteins without a known crystal structure to be more accurately modeled than other comparative modeling programs. This is due to the incorporation of experimental data as spatial restraints imposed onto the target sequence. According to Lesk, protein sequence determines protein structure and this in turn determines protein function.^[12] Thus, comparative models allow for functional studies and annotation of proteins by fold recognition. Research that have been carried out based on comparative models include:

• Molecular replacement in X-ray crystallography^[13]
• Effect of Single-Nucleotide Polymorphisms on human diseases^[14]

Comparative modeling with MODELLER was also adpated in 3D-Jury, a meta-predictor of protein structures.^[15]

Some Notable Alumni from Šali Lab

Andras Fiser

• Number of Publications: 77
• Average Citation per Item: 53.32
• Number of Times Cited: 4106
• Average Citation per Year: 165.79
• H-index: 25

Marc A. Marti-Renom

• Number of Publications: 56
• Average Citation per Item: 45.25
• Number of Times Cited: 2534
• Average Citation per Year: 194.92
• H-index: 19

M.S. Madhusudhan

• Number of Publications: 35
• Average Citation per Item: 32.14
• Number of Times Cited: 1125
• Average Citation per Year: 86.54
• H-index: 15

References

1. ^ Andrej Šali’s Publications
2. ^ Web of Science
3. ^ SALI, A., SHAKHNOVICH, E. & KARPLUS, M. 1994. HOW DOES A PROTEIN FOLD. Nature, 369, 248-251.
4. ^ MARTI-RENOM, M. A., STUART, A. C., FISER, A., SANCHEZ, R., MELO, F. & SALI, A. 2000. Comparative protein structure modeling of genes and genomes. Annual Review of Biophysics and Biomolecular Structure, 29, 291-325.
5. ^ BAKER, D. & SALI, A. 2001. Protein structure prediction and structural genomics. Science, 294, 93-96.
6. ^ ROHL, C. A., STRAUSS, C. E. M., CHIVIAN, D. & BAKER, D. 2004. Modeling structurally variable regions in homologous proteins with rosetta. Proteins-Structure Function and Bioinformatics, 55, 656-677.
7. ^ FISER, A., DO, R. K. G. & SALI, A. 2000. Modeling of loops in protein structures. Protein Science, 9, 1753-1773.
8. ^ SALI, A. & BLUNDELL, T. L. 1993. COMPARATIVE PROTEIN MODELING BY SATISFACTION OF SPATIAL RESTRAINTS. Journal of Molecular Biology, 234, 779-815.
9. ^ MODELLER Manual
10. ^ MODELLER Manual
11. ^ MODELLER Manual
12. ^ Chothia, C. and Lesk, A.M. The relation between the divergence of sequence and structure in proteins. EMBO J. (1986), 5(4), 823-826.
13. ^ KOEPKE, J., HU, X. C., MUENKE, C., SCHULTEN, K. & MICHEL, H. 1996. The crystal structure of the light-harvesting complex II (B800-850) from Rhodospirillum molischianum. Structure, 4, 581-597.
14. ^ CHASMAN, D. & ADAMS, R. M. 2001. Predicting the functional consequences of non-synonymous single nucleotide polymorphisms: Structure-based assessment of amino acid variation. Journal of Molecular Biology, 307, 683-706.
15. ^ GINALSKI, K., ELOFSSON, A., FISCHER, D. & RYCHLEWSKI, L. 2003. 3D-Jury: a simple approach to improve protein structure predictions. Bioinformatics, 19, 1015-1018.

External links

• Homepage of Šali's laboratory
• Explanation at Business 2.0