SSM vs. others: 1GTH: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
 
1GTH:A (1019 residues)
DIHYDROPYRIMIDINE DEHYDROGENASE (DPD) FROM PIG,
TERNARY COMPLEX WITH NADPH AND 5-IODOURACIL

43 longest helices and 31 strands were used for SSE matching.
 

1.  V A S T    (server)

Figure 1GTH: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 structural neighbours may be classified into highly similar (PDB entries 1-20 in the Figure) and very remote ones. There are no PDB entries with 30% to 98% residues aligned to the input 1GTH:A, and both SSM and VAST agree on that. The SSM curve look like a fit to that of VAST in Figure 1GTH:A-1, which indicates a generally good agreement. Yet, for individual entries there is a 50%-discreapancy in the number of aligned residues.


  Figure 1GTH: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.
 

Figure 1GTH:A-2 suggests that SSM offers alignments with a somewhat lower RMSDs, on comparison with VAST. Both SSM and VAST produce alignments with RMSDs in a fairly reasonable range (less than 5 Å) and generally agree on the corridor of RMSD values.


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

Visual inspection of Figures 1GTH:A-1 and 1GTH: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 1GTH:A-3, is slightly higher than that calculated from VAST results, although the difference is not large. A detail analysis of the Figure shows that both SSM and VAST curves replicate each other very closely down to small local details.


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

SSM gives lower P-values for highly similar structures and comes closer to numerical agreement with VAST in the region of remote structural neighbours (cf. Figure 1GTH:A-4). The general trend (increasing P-values with decreasing similarity) is there in both SSM and VAST results, however high dissimilarity of structures correponding to PDB entries 21 and higher in the Figure, does not allow to conclude about agreement between SSM and VAST more specifically.


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

Z-scores of Ca-alignments, given by SSM, are almost exactly twice lower than those obtained from VAST for remote structures (Figure 1GTH:A-5). For highly similar structures (PDB entries 1-20), SSM's Z-scores are 3 times lower than those from VAST. The general trend of decreasing Z-scores with decreasing structural similarity is there in both results. In spite of numerical difference, both curves in Figure 1GTH:A-5 show similarity in relative changes; this allows one to rate the agreement between SSM and VAST Z-scores as acceptable.


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

 

 

2.  C E (Combinatorial Extension)    (server)

CE gives a very much the same picture as VAST and SSM (cf. Figures 1GTH:A-6 and 1GTH:A-1) however offers more matches. Just as VAST, CE does not find Ca-alignments in the range of 30%-98% of aligned input residues. SSM seems to be giving, on average, longer alignments in the region of PDB entries 150-800, which is not typical for the comparison with CE. Most of alignment lengths, obtained from SSM and CE, agree within 10% of difference, yet in some instances the difference reaches 40% or so.


  Figure 1GTH: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 1GTH:A-7 shows that SSM produces alignments with shorter, on average, RMSDs, as comapred to those from CE. This holds true even for the region in which SSM alignments are longer (PDB entries 150-800). Both servers give a reasonable range of RMSDs (less than 5 Å).


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

The match indexes, calculated from SSM and CE results, indicates a slightly better quality of SSM alignments for individual structures, however the difference between SSM and CE is not particularly big (cf. Figure 1GTH:A-8). A detail analysis of the Figure yields that SSM and CE alignments have indeed a very close match indexes, which is an indication of a good agreement in principal quality of alignment. The differences in alignment lengths and RMSDs, visible in Figures 1GTH:A-6 and 1GTH:A-7 should therefore be attributed to the difference in balancing the compromise between them, employed by the servers. Usually, CE favours longer alignments at higher RMSDs, however in the particular example of 1GTH:A one cannot probably say so. Just as RMSDs and alignment lengths (cf. Figures 1GTH:A-6 and 1GTH:A-7), match index shows a clear distinction between highly similar and remote structural neighbours.


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

With the exception for the highly similar structures, Z-scores from SSM agree reasonably well with those from CE, after applying a factor of 2 to the latter (which we do in all comparisons, cf. Figure 1GTH:A-9). For highly similar structures, SSM gives a considerably higher Z-score, making a much more pronounced distinction between similar and dissimilar structures. Also in difference of CE, SSM's Z-scores go down for the most remote structures, showing a good correlation with data presented in Figures 1GTH:A-6, 1GTH:A-7 and 1GTH:A-8.


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

 

 

3.  D A L I    (server)

DALI fails to recognize 1GTH:A as an input structure, and does not find any of its closest structural neighbours (cf. Figure 1GTH:A-10). On comparison with VAST and CE, DALI returns much less hits, all of them relate to remote structures (less than 30% of input's residues aligned). SSM agrees with DALI reasonably well, the average deviation in the lengths of Ca-alignments makes about 15%, reaching 40% in some instances. SSM results are well centered in respect to those from DALI: as may be seen from Figure 1GTH:A-10, SSM curve looks almost as a fit to that of DALI.


  Figure 1GTH: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.
 

Figure 1GTH:A-11 shows a considerable difference in RMSDs of SSM and DALI alignments. As seen from the Figure, most of RMSDs from DALI are significantly (severl times) higher than those given by SSM, with only 4 exceptions, where DALI's RMSDs are slightly lower than SSM ones. The highest RMSD, reported by DALI, makes more than 16 Å for PDB entry 1REQ:A (SSM gives 3.64 Å). Given a general agreement in alignment lengths (cf. Figure 1GTH:A-11), one should conclude that SSM produces alignments of a higher quality.


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

Indeed, the match index, calculated from SSM results, is noticeably higher than that obtained from DALI's output (cf. Figure 1GTH:A-12). Higher match index generally indicates a higher quality of 3D alignment (longer alignments at same RMSD), therefore one can conclude that for most SSM alignments, presented in the Figure, have a better quality on comparison with those from DALI. This difference is not typical for the comparison with DALI; on contrary, in most cases the principal quality of SSM and DALI alignments, as meaured by match index, is found nearly identical.


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

Z-scores from SSM and DALI do not show an evident similarity (cf. Figure 1GTH:A-13), which is usually observed for remote structural neihgbours. SSM Z-scores are, on average, higher than those from DALI and look less scattered. We always compare DALI Z-scores with the minus logarithm of SSM's P-values (black line in Figure 1GTH:A-13); in the present case, they hardly show any similarity.


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

 

 

4.  D E J A V U    (server)

Similarly to DALI, DEJAVU does not recognize the input structure and does not find any of its closest structural neighbours. As seen from Figure 1GTH:A-14, DEJAVU gives matches with less than 30% (most of them with less than 16%) of the input residues aligned. Typically for the comparison with DEJAVU, SSM produces noticeably longer Ca-alignments; most of SSM alignments are 30% longer than those obtained from DEJAVU.


  Figure 1GTH: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.
 

Figure 1GTH:A-15 makes it clear that longer SSM alignments come at the expense of higher RMSDs. DEJAVU produces considerably (by 60% on average) 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.


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

Match indexes, shown in Figure 1GTH:A-16, indicate the balance between alignment length and RMSD. Higher match indexes correspond to longer alignments at same RMSD, and therefore to higher quality of alignment. As seen from the Figure, SSM offers, on average, alignments of a higher quality than those produced by DEJAVU. Although there are examples of a better quality of DEJAVU alignments as well, it seems that SSM prevails in larger number of instances. Analysis of Figures 1GTH:A-14, 1GTH:A-15 and 1GTH:A-16 suggests that at similar quality of 3D alignments, SSM and DEJAVU differ significantly in balancing the alignment length and RMSD.


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

P-values from DEJAVU are noticeably lower than those given by SSM (cf. Figure 1GTH:A-17). In a few cases, DEJAVU gives zero P-values, which we find somewhat oddish. Zero P-value means that there is no chance to get match of a better quality, that is, by definition, matching a structure to itself. Therefore assigning zero P-values to matches with evidently remote structural neighbours looks confusing.


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

Z-score is, generally, a reflection of P-value. As seen in Figure 1GTH:A-18, DEJAVU tends to much higher Z-scores on comparison with SSM. Although a very general trend in SSM and DEJAVU Z-scoring may be identified as similar (decreasing Z-score with increasing dissimlarity), the overall agreement can be hardly rated as a good one.


  Figure 1GTH:A-18.
Z-scores corresponding to matches shown in Figure 1GTH: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 VAST and CE. At the same time SSM makes alignments of better, on average, quality than obtained from DALI and DEJAVU. Different servers differ in solving the compromise between alignment length in RMSD. In that respect, SSM agrees reasonably well with VAST and CE, shows a poorer agreement with DALI and 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 and DEJAVU fail to identify the input structure and all closest structural neighbours. The difference in alignment lengths between SSM and DEJAVU is significant. The corresponding RMSDs from VAST and CE are close to those offered by SSM. DALI produces considerably higher RMSDs, as compared to SSM, resulting, at comparable alignment lengths, in noticeably lower alignment quality, as indicated by match index. RMSDs from DEJAVU are considerably lower than those from SSM, however corresponding to them match indexes show poorer quality of alignment for many PDB entries. SSM agrees reasonably well with P-values and Z-scores from VAST and CE. DALI and DEJAVU show only very general agreement with SSM in Z-scoring (and P-scoring for DEJAVU), which is probably due to the fact that these servers miss all closest structural neighbours.