Limits...
An automatic method for assessing structural importance of amino acid positions.

Sadowski MI, Jones DT - BMC Struct. Biol. (2009)

Bottom Line: Some buried residues were found to be less informative than expected, particularly residues involved in active sites and the binding of small molecules.The correlation-based method generates predictions of structural importance for superfamily positions which agree well with previous results of manual analyses, and may be of use in automated residue annotation piplines.A PERL script which implements the method is provided.

View Article: PubMed Central - HTML - PubMed

Affiliation: Computer Science Department, University College London, Gower St, London, WC1E 6BT, UK. msadows@nimr.mrc.ac.uk

ABSTRACT

Background: A great deal is known about the qualitative aspects of the sequence-structure relationship, for example that buried residues are usually more conserved between structurally similar homologues, but no attempts have been made to quantitate the relationship between evolutionary conservation at a sequence position and change to global tertiary structure. In this paper we demonstrate that the Spearman correlation between sequence and structural change is suitable for this purpose.

Results: Buried residues, bends, cysteines, prolines and leucines were significantly more likely to occupy positions highly correlated with structural change than expected by chance. Some buried residues were found to be less informative than expected, particularly residues involved in active sites and the binding of small molecules.

Conclusion: The correlation-based method generates predictions of structural importance for superfamily positions which agree well with previous results of manual analyses, and may be of use in automated residue annotation piplines. A PERL script which implements the method is provided.

Show MeSH

Related in: MedlinePlus

Distributions of positional rank-correlations with global structural change for six superfamilies. Histograms of Spearman's rank correlation coefficients with bin widths of 0.05 for the six datasets (dashed black lines). Partial correlations controlled for global sequence-structure similarity are also plotted (solid red lines). Datasets are ordered left-to-right: amylase, cupredoxin, globin, lysozyme, viral capsid (jellyroll), phospholipase A2 (snake venom toxin).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2667441&req=5

Figure 1: Distributions of positional rank-correlations with global structural change for six superfamilies. Histograms of Spearman's rank correlation coefficients with bin widths of 0.05 for the six datasets (dashed black lines). Partial correlations controlled for global sequence-structure similarity are also plotted (solid red lines). Datasets are ordered left-to-right: amylase, cupredoxin, globin, lysozyme, viral capsid (jellyroll), phospholipase A2 (snake venom toxin).

Mentions: Figure 1 depicts the distribution of rank-correlation coefficients for each of the six superfamilies (red, dashed lines). In order to account for the overall sequence/structure relationship the distributions of partial correlations which account for the relationship to global sequence similarity are also shown (black lines). We use the partial correlation measure in later analyses since it removes the background noise from sequence similarity.


An automatic method for assessing structural importance of amino acid positions.

Sadowski MI, Jones DT - BMC Struct. Biol. (2009)

Distributions of positional rank-correlations with global structural change for six superfamilies. Histograms of Spearman's rank correlation coefficients with bin widths of 0.05 for the six datasets (dashed black lines). Partial correlations controlled for global sequence-structure similarity are also plotted (solid red lines). Datasets are ordered left-to-right: amylase, cupredoxin, globin, lysozyme, viral capsid (jellyroll), phospholipase A2 (snake venom toxin).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2667441&req=5

Figure 1: Distributions of positional rank-correlations with global structural change for six superfamilies. Histograms of Spearman's rank correlation coefficients with bin widths of 0.05 for the six datasets (dashed black lines). Partial correlations controlled for global sequence-structure similarity are also plotted (solid red lines). Datasets are ordered left-to-right: amylase, cupredoxin, globin, lysozyme, viral capsid (jellyroll), phospholipase A2 (snake venom toxin).
Mentions: Figure 1 depicts the distribution of rank-correlation coefficients for each of the six superfamilies (red, dashed lines). In order to account for the overall sequence/structure relationship the distributions of partial correlations which account for the relationship to global sequence similarity are also shown (black lines). We use the partial correlation measure in later analyses since it removes the background noise from sequence similarity.

Bottom Line: Some buried residues were found to be less informative than expected, particularly residues involved in active sites and the binding of small molecules.The correlation-based method generates predictions of structural importance for superfamily positions which agree well with previous results of manual analyses, and may be of use in automated residue annotation piplines.A PERL script which implements the method is provided.

View Article: PubMed Central - HTML - PubMed

Affiliation: Computer Science Department, University College London, Gower St, London, WC1E 6BT, UK. msadows@nimr.mrc.ac.uk

ABSTRACT

Background: A great deal is known about the qualitative aspects of the sequence-structure relationship, for example that buried residues are usually more conserved between structurally similar homologues, but no attempts have been made to quantitate the relationship between evolutionary conservation at a sequence position and change to global tertiary structure. In this paper we demonstrate that the Spearman correlation between sequence and structural change is suitable for this purpose.

Results: Buried residues, bends, cysteines, prolines and leucines were significantly more likely to occupy positions highly correlated with structural change than expected by chance. Some buried residues were found to be less informative than expected, particularly residues involved in active sites and the binding of small molecules.

Conclusion: The correlation-based method generates predictions of structural importance for superfamily positions which agree well with previous results of manual analyses, and may be of use in automated residue annotation piplines. A PERL script which implements the method is provided.

Show MeSH
Related in: MedlinePlus