SSM vs. others: 1FIN: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
 
1FIN:A (298 residues)
CYCLIN A - CYCLIN-DEPENDENT KINASE 2 COMPLEX
12 longest helices and 7 strands were used for SSE matching.
 

1.  V A S T    (server)

Figure 1FIN: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. Ca-alignments agree better for more similar structures, while for more remote ones SSM offers somewhat shorter alignments, which is not typical for comparison with VAST.


  Figure 1FIN: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.
 

SSM produces alignments with lower RMSDs, as may be seen from Figure 1FIN:A-2. Both SSM and VAST produce alignments with RMSDs in a fairly reasonable range (less than 5 Å).


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

Analysis of Figures 1FIN:A-1 and 1FIN:A-2 suggests that SSM may produce alignments of a better quality than given by VAST (lower RMSDs at similar alignment lengths). Indeed, SSM's match index, shown by black line in Figure 1FIN:A-3, is slightly higher than that calculated from VAST results. It is interesting to note that both SSM and VAST curves in Figure 1FIN:A-3 behave in a very much the same way, demonstrating similarity in quite small details.


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

SSM gives lower P-values for more similar structures and comes closer to numerical agreement with VAST in the region of remote structural neighbours (cf. Figure 1FIN:A-4). The general trend (increasing P-values with decreasing similarity) is there in both SSM and VAST results, however SSM gives a more pronounced dependence.


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

Z-scores of Ca-alignments, given by SSM, are noticeably lower than those obtained from VAST (Figure 1FIN:A-5). The general trend of decreasing Z-scores with decreasing structural similarity is there in both results, however the dependence is rather poor.


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

 

 

2.  C E (Combinatorial Extension)    (server)

Ca-alignments, obtained from CE, allow for classification to close (PDB entries 1-138), less similar (139-155) and remote (156-209) structural neighbours (cf. Figure 1FIN:A-6). These regions of different similarity are clearly identifiable in the Figure by sharp changes in the alignment length. As seen from the Figure, SSM reproduces this classification, giving lower, on average, alignment lengths for close structures and showing a better agreement for less similar and remote ones. SSM and CE give exactly the same borderlines between the regions of different similarity.


  Figure 1FIN: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.
 

As demonstrated in Figure 1FIN:A-7, SSM produces alignments with shorter, on average, RMSDs, as comapred to those from CE. Both servers give a reasonable range of RMSD (less than 5 Å). It may be derived from comparison of Figures 1FIN:A-6 and 1FIN:A-7 that CE makes its alignments longer solely for the expense of higher RMSDs. The borderline between similar and remote structural neighbours at PDB entry 139 is clearly seen in both SSM and CE results, however the intermediate level of similarity (PDB entries 139-155, cf. Figure 1FIN:A-6) is not identifiable in RMSDs.


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

The above conclusion is confirmed by the match index, shown in Figure 1FIN:A-8. As seen from the Figure, SSM and CE alignments have very close match indexes, which is an indication of a good agreement in principal quality of alignment. The differences between the results shown in Figures 1FIN:A-6 and 1FIN:A-7 are therefore a consequence of how the servers balance the compromise between alignment length and RMSD. Evidently, CE favours longer alignments at higher RMSDs. Note that quality of SSM alignments, as measured by match index, is slightly higher for more similar structures, on comparison with CE. Just as RMSDs and alignment lengths (cf. Figures 1FIN:A-6 and 1FIN:A-7), match index shows a clear borderline between similar and remote structural neighbours at PDB entry 139.


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

Figure 1FIN:A-9 shows that Z-scores, given by SSM, agree reasonably well with those from CE, after applying a factor of 2 to the latter (which we do in all comparisons). Z-scores from both servers allow for the classification of structural neighbours by their similarity to 1FIN:A, as clearly seen in Figure 1FIN:A-6 with the remark that the intermediate similarity level (PDB entries 139-155) is not well identified in Z-scores.


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

 

 

3.  D A L I    (server)

DALI recognizes 1FIN:A as an input structure, but, as may be seen from Figure 1FIN:A-10, fails to align all its residues. The same is observed for the most similar structural neighbours (PDB entries 1-12 in the picture), however apart of that, the overall agreement between DALI and SSM may be rated as good. SSM offers systematically shorter alignments for less similar structural neighbours (at about 5% difference, PDB entries 42-84), but gives a nice average of alignment lengths from DALI for remote structures. The servers give the same borderline between PDB entries with high and low similarity to 1FIN:A.


  Figure 1FIN: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.
 

In spite of not aligning all residues of the input structure to itself, DALI reports zero RMSD for that alignment, as seen from Figure 1FIN:A-11, PDB entry 1. This may indicate a minor problem with DALI's database. The Figure demonstrates that SSM and DALI produce similar RMSDs, yet RMSDs from SSM are visibly lower than DALI's in almost all instances. For one particular structure, 1JVP:P, DALI gives an unreasonably high RMSD, which is definitely wrong. With this single exception, both SSM and DALI RMSD curves in Figure 1FIN:A-11 agree very well in nearly all small details.


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

Comparison of Figures 1FIN:A-10 and 1FIN:A-11 suggests that SSM and DALI may agree in the principal quality of alignment: SSM's alignments are shorter at shorter RMSDs. However, match index calculated from SSM results is somewhat higher, on average, than that of DALI (cf. Figure 1FIN:A-12), indicating a better quality of SSM alignments. The RMSD spike for PDB entry 1JVP:P, dominating Figure 1FIN:A-11, is reflected in sharp decrease in match quality for that entry. As seen from Figure 1FIN:A-12, relative changes in both SSM and DALI match index curves follow each other very nicely, which shows a general agreement between the alignment algorithms. The differences in alignment lengths and RMSDs, seen in Figures 1FIN:A-10 and 1FIN:A-11 should mean that SSM and DALI differ, albeit not significantly, in balancing the alignment length and RMSD.


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

Z-scores, given by SSM, are lower for more similar structural neighbours and higher for the remote ones, as compared to DALI (cf. Figure 1FIN:A-13). Typically for comparison with DALI, Z-scores from DALI agree better with the minus logarithm of SSM's P-values (black line in Figure 1FIN:A-13).


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

 

 

4.  D E J A V U    (server)

DEJAVU failed to recognize the input and gave 1IA8 as its closest prototype. While VAST gives 182 close structural neighbours, 155 are found in CE output, and DALI lists 84 of them, DEJAVU finds just 3: 1IA8, 1QPC and 1FMK. Most of DEJAVU output relates to very remote structural neighbours, with less than 17% of input's residues aligned (30% aligned in SSM results) (cf. Figure 1FIN:A-14). With the exception for just a few structures, SSM produces considerably longer Ca-alignments, as compared to DEJAVU.


  Figure 1FIN: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 Figure 1FIN:A-15 shows, longer alignments from SSM come at the expense of higher RMSDs. DEJAVU produces considerably shorter RMSDs, as compared to SSM, for almost all structures. While DEJAVU evidently keeps RMSD in the range below 2.25 Å, SSM allows for RMSD up to 5 Å and even more in couple of instances corresponding to a very dissimilar structures, for which alignment itself is probably a non-existing notion (just 16 aligned residues out of 298).


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

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


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

P-values from DEJAVU are noticeably lower than those obtained from SSM (cf. Figure 1FIN: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 1FIN:A-17.
P-values corresponding to matches shown in Figure 1FIN: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 1FIN:A-18). Agreement in Z-scores is probably slightly better than that seen in P-values.


  Figure 1FIN:A-18.
Z-scores corresponding to matches shown in Figure 1FIN: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. Yet, SSM makes better-quality alignments than those from VAST and DALI, as indicated by the match index, although the difference is not high. Different servers differ 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 close to those obtained from VAST, CE and DALI, meaning that the average deviation of the alignment lengths is sufficiently low. DALI fails to align all input's residues to the most similar structural neighbours, including the input itself. The difference in alignment lengths 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 corresponding to them 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.