SSM vs. others: 1LDC:A

 
Materials from this page cannot be reproduced without permission from the authors.
Comparisons made on November 2002, using current versions of VAST, CE, DALI, DEJAVU and SSM v1.22 from 20/11/2002.
 

CONTENTS
  1. VAST
  2. CE (Combinatorial Extension)
  3. DALI
  4. DEJAVU
  5. Conclusion
 
1LDC:A (478 residues)
L-LACTATE DEHYDROGENASE: CYTOCHROME C OXIDOREDUCTASE
21 longest helices and 10 strands were used for SSE matching.
 

1.  V A S T    (server)

Figure 1LDC:A-1 shows the Ca-alignment lengths obtained from SSM and VAST for different structural neighbours (as chosen by VAST). As seen from the picture, SSM and VAST produce similar results. The difference in alignment lengths is less then 10% in most cases, however there are a few cases where the difference reaches 50%. Ca-alignments given by SSM are slightly longer, on average, than those obtained from VAST.


  Figure 1LDC:A-1.
Length of Ca-alignment as a function of PDB entry, obtained by SSM (black line) and VAST (red line). Details of the calculations are given here.
 

The root mean square deviations (RMSDs) show roughly the same degree of similarity as the alignment lengths (cf. Figure 1LDC:A-2). Both SSM and VAST produce alignments with RMSDs in a fairly reasonable range (less than 5 Å). It may be derived from the picture, that SSM produces slightly lower RMSDs, as compared to VAST.


  Figure 1LDC:A-2.
RMSD of Ca-alignment corresponding to data in Figure 1LDC:A-1. Details of the calculations are given here.
 

Analysis of Figures 1LDC:A-1 and 1LDC:A-2 suggests that SSM and VAST produce a similar quality alignments. Indeed, match index shown in Figure 1LDC:A-3 shows a high degree of similarity. The match index, calculated from SSM results, is slightly higher than that corresponding to VAST alignments, indicating a somewhat better quality of SSM alignments; this is a direct consequence of longer, on average, SSM alignments and shorter, also on average, RMSDs given by SSM. It may be derived from Figures 1LDC:A-1, 1LDC:A-1 and 1LDC:A-3 that SSM and VAST agree well in the principal quality of 3D alignments, however differ in balancing between alignment length and RMSD.


  Figure 1LDC:A-3.
Match Index corresponding to data shown in Figure 1LDC:A-1. Details of the calculations are given here.
 

P-values of SSM and VAST alignments, shown in Figure 1LDC:A-4, demonstrate similar trends and run in approximately the same corridor of values, but do not show a full agreement. As seen from the Figure, SSM assignes a higher statistical significance to highly similar structural neighbours, but underestimates the remote ones, as compared to VAST.


  Figure 1LDC:A-4.
P-values corresponding to matches shown in Figure 1LDC:A-1. Details of the calculations are given here.
 

Z-scores of Ca-alignments, given by VAST, are twice higher than those from SSM (Figure 1LDC:A-5). Apart from that, SSM and VAST demonstrate a relatively good agreement in Z-scoring. Indeed, SSM and VAST curves in Figure 1LDC:A-5 show correlation in many details, including most positive and negative spikes.


  Figure 1LDC:A-5.
Z-scores corresponding to matches shown in Figure 1LDC:A-1. Details of the calculations are given here.
 

 

 

2.  C E (Combinatorial Extension)    (server)

Judging by the lengths of Ca-alignments, SSM and CE results (Figure 1LDC:A-6) show approximatly the same degree of similarity as that seen between SSM and VAST (cf. Figure 1LDC:A-1). In difference of VAST, CE shows two groups of remote structural neighbours, with a graduate transition at PDB entries 1000-1150 between them. As seen from the Figure, SSM reproduces this particular feature. Since both Figures 1LDC:A-1 and 1LDC:A-6 are ordered by decreasing length of SSM alignments, one concludes that VAST and CE differ in choosing the structural neighbours. Comparison of alignment lengths suggests that most VAST results correspond to less remote structural neighbours in Figure 1LDC:A-6, while most of PDB entries 1100-1900 from that picture are omitted by VAST (due to their insignificance). As seen from the Figure 1LDC:A-6, alignment lengths from SSM are well centered against those produced by CE, with the exception of the most remote structures.


  Figure 1LDC:A-6.
Length of Ca-alignment as a function of PDB entry, obtained by SSM (black line) and CE (red line). Details of the calculations are given here.
 

Figure 1LDC:A-7 shows that SSM produces shorter, on average, RMSDs of Ca-alignment, as comapred to CE. RMSDs from SSM stay below 5 Å but show the same range of variations in absolute value as those obtained from CE. Comparing Figures 1LDC:A-6 and 1LDC:A-7 one may suggest that SSM gives a somewhat shorter RMSDs at similar alignment lengths.


  Figure 1LDC:A-7.
RMSD of Ca-alignment corresponding to data in Figure 1LDC:A-6. Details of the calculations are given here.
 

Indeed, the match index, indicating the principal quality of 3D alignment, is somewhat higher for SSM alignments (cf. Figure 1LDC:A-8). This feature is noticeable but not attributable to all structures represented in the Figure. Analysis of Figures 1LDC:A-6, 1LDC:A-7 and 1LDC:A-8 suggests that SSM and CE generally agree in the principal quality of the alignment, but differ in balancing the compromise between alignment length and RMSD.


  Figure 1LDC:A-8.
Match Index corresponding to data shown in Figure 1LDC:A-6. Details of the calculations are given here.
 

Z-scores of SSM are on average twice higher than those obtained from CE (cf. Figure 1LDC:A-9). Comparison of Figures 1LDC:A-8 and 1LDC:A-9 indicates that Z-scores from SSM are better correlated with the quality of alignment, given by match index than what is given by CE. SSM Z-scores give also a better indication of less and more remote structural neighbours, which are identifiable in Figure 1LDC:A-6


  Figure 1LDC:A-9.
Z-scores corresponding to matches shown in Figure 1LDC:A-6. Details of the calculations are given here.
 

 

 

3.  D A L I    (server)

Comparing to VAST and CE, DALI produces considerably less matches (cf. Figure 1LDC:A-10). As seen from the Figure, DALI fails to align all residues for the most similar structures (PDB entries 1-12), reporting 327 aligned residues of total 478 for 1LDC:A itself. Except this difference, SSM agrees with DALI very well. The observed difference in the alignment lengths is about 5% on average, and SSM results look almost as a fit to those of DALI.


  Figure 1LDC:A-10.
Length of Ca-alignment as a function of PDB entry, obtained by SSM (black line) and DALI (red line). Details of the calculations are given here.
 

Having aligned some 68% of the input's residues to 1LDC:A itself, DALI reports zero RMSD for that alignment, as seen from Figure 1LDC:A-11, PDB entry 1. The Figure demonstrates that SSM and DALI produce similar RMSDs, yet RMSDs from SSM are lower than DALI's for remote structural neighbours. It is also seen that there is a good agreement in small details of both RMSD curves.


  Figure 1LDC:A-11.
RMSD of Ca-alignment corresponding to data in Figure 1LDC:A-10. Details of the calculations are given here.
 

The match indexes, calculated from SSM and DALI alignments, show a remarkable agreement (Figure 1LDC:A-12). This does not apply to the most similar structures, which are not fully aligned by DALI. For all other structures, SSM and DALI curves are nearly coinciding. This result indicates that the servers give the same principal quality of alignment. The differences in alignment lengths and RMSDs, seen in Figures 1LDC:A-10 and 1LDC:A-11 then mean that SSM and DALI differ, albeit not significantly, in balancing the alignment length and RMSD.


  Figure 1LDC:A-12.
Match Index corresponding to data shown in Figure 1LDC:A-10. Details of the calculations are given here.
 

SSM gives lower Z-score for most similar structural neighbours, as compared to DALI, while Z-scores for remote structures agree reasonably well (Figure 1LDC:A-13). Typically for comparison with DALI, DALI's Z-scores for highly similar structures agree better with minus logarithm of SSM's P-values (black line in Figure 1LDC:A-13).


  Figure 1LDC:A-13.
Z-scores corresponding to matches shown in Figure 1LDC:A-10. Details of the calculations are given here.
 

 

 

4.  D E J A V U    (server)

DEJAVU failed to recognize most closest structural neighbours, including 1LDC:A itself and gave 1D3G as the closest prototype of the input (cf. Figure 1LDC:A-14). As seen from the Figure, this PDB entry should be rated as less similar structural neighbour to 1LDC:A as only 50% of its residues were aligned by SSM (and 42% by DEJAVU). Most of DEJAVU output represents very remote structures with less than 25% of input's residues aligned. For all but few structures, SSM produces considerably longer Ca-alignments, as compared to DEJAVU.


  Figure 1LDC:A-14.
Length of Ca-alignment as a function of PDB entry, obtained by SSM (black line) and DEJAVU (red line). Details of the calculations are given here.
 

As seen from Figure 1LDC:A-15, longer alignments from SSM come at the expense of higher RMSDs. DEJAVU produces considerably shorter RMSDs, as compared to SSM. While DEJAVU evidently keeps RMSD in the range below 2.5 Å, SSM allows for RMSD up to 5 Å and even more in one instance.


  Figure 1LDC:A-15.
RMSD of Ca-alignment corresponding to data in Figure 1LDC:A-14. Details of the calculations are given here.
 

The balance between alignment length and RMSD is indicated by match indexex. As seen from Figure 1LDC:A-16, match indexes from SSM and DEJAVU have similar values, with SSM indexes probably somewhat higher on average. Analysis of Figures 1LDC:A-14, 1LDC:A-15 and 1LDC:A-16 suggest that at similar quality of 3D alignments, SSM and DEJAVU differ significantly in balancing the alignment length and RMSD.


  Figure 1LDC:A-16.
Match Index corresponding to data shown in Figure 1LDC:A-14. Details of the calculations are given here.
 

P-values from DEJAVU are noticeably lower than those obtained from SSM (cf. Figure 1LDC:A-17). Although the Figure shows some correlation between the SSM and DEJAVU curves, the overall agreement in P-values should be rated as poor.


  Figure 1LDC:A-17.
P-values corresponding to matches shown in Figure 1LDC:A-14. Details of the calculations are given here.
 

In accordance with P-values, Z-scores from DEJAVU are higher than those given by SSM (Figure 1LDC:A-18). Agreement in Z-scores is probably slightly better than that seen in P-values.


  Figure 1LDC:A-18.
Z-scores corresponding to matches shown in Figure 1LDC:A-14. Details of the calculations are given here.
 

 

 

5.  Conclusion

The principal quality of 3D Ca-alignments, as measured by match index, is in a good agreement between results produced by SSM and those obtained from other servers. The comparison is particularly good between SSM and DALI. However, the servers show a difference in solving the compromise between alignment length in RMSD. In that respect, SSM agrees well with VAST, CE, DALI, while shows a considerable difference from DEJAVU.

SSM alignments are relatively well centered to those obtained from VAST, CE and DALI, meaning that the average deviation of the alignment lengths is sufficiently low. DALI fails to properly align the most similar structural neighbours, including the input one. The difference in alignment length between SSM and DEJAVU is significant. The corresponding RMSDs from VAST, CE and DALI are close to those offered by SSM, although SSM seem to produce slightly better alignments with lower RMSDs at comparable alignment lengths (resulting in a slightly higher match index). RMSDs from DEJAVU are considerably lower than those from SSM, however their match indexes show similar quality of alignment. SSM agrees reasonably well with P-values and Z-scores from VAST, CE and DALI. This allows one to use them for rating the matches, expecting that the correspondence in the rating will be preserved, on average, across different servers.