SSM vs. others: 4DFR: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
 
4DFR:A (160 residues)
DIHYDROFOLATE REDUCTASE (E.C.1.5.1.3) COMPLEX
4 longest helices and 8 strands were used for SSE matching.
 

1.  V A S T    (server)

The picture below (Figure 4DFR:A-1) shows comparison of Ca-alignment length obtained from SSM and VAST. As seen from the picture, both SSM and VAST give clearly distinguished sets of close (PDB entries 1-118) and remote (PDB entries 119 and higher) structural neighbours.


  Figure 4DFR: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 demonstrates a very close agreement with VAST for highly similar structures, and offers somewhat longer Ca-alignments for more remote structural neighbours. As may be clarified from Figure 4DFR:A-2, the latter naturally comes in exchange for larger RMSDs, which however remain in a fairly reasonable range.


  Figure 4DFR:A-2.
RMSD of Ca-alignment corresponding to data in Figure 4DFR:A-1. Details of the calculations are given here.
 

Analysis of Figures 4DFR:A-1 and 4DFR:A-2 shows that SSM and VAST produce very similar results balanced slightly differently in respect to the alignment length and RMSD. The latter is confirmed by data shown in Figure 4DFR:A-3: it appears that match index of SSM alignments is a very close replica of that derived from VAST results.


  Figure 4DFR:A-3.
Match Index corresponding to data shown in Figure 4DFR:A-1. Details of the calculations are given here.
 

P-values of SSM and VAST alignments, shown in Figure 4DFR:A-4, demonstrate similar trends and allow for clear identification of highly similar, similar and remote structural neighbours. SSM gives lower P-values as compared to those of VAST, especially for highly similar neighbours.


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

Just as P-values, Z-scores of SSM and VAST (Figure 4DFR:A-5) show a reasonably good qualitative agreement, although differ in numbers.


  Figure 4DFR:A-5.
Z-scores corresponding to matches shown in Figure 4DFR:A-1. Details of the calculations are given here.
 

 

 

2.  C E (Combinatorial Extension)    (server)

Unlike VAST, Combinatorial Extension makes noticeably longer Ca-alignments, as compared to SSM (cf. Figure 4DFR:A-6). General trends of SSM and CE curves in Figure 4DFR:A-6 agree very well (note that the results are ordered by decreasing the lengths of Ca-alignments given by SSM), and allow for the identification of highly similar (entries 1-60), similar (61-110) and remote ( 111-452) structural neighbours.


  Figure 4DFR: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 4DFR:A-7 shows that CE produces, on average, Ca-alignments with RMSDs larger than those from SSM. The overall correlation between alignment length and RMSD (the larger is one, the less is the other) is good, as may be found from the analysis of Figures 4DFR:A-6 and 4DFR:A-7.


  Figure 4DFR:A-7.
RMSD of Ca-alignment corresponding to data in Figure 4DFR:A-6. Details of the calculations are given here.
 

The match index of SSM and CE alignments shows a very good agreement. It is interesting to see that pronounced differences between SSM and CE alignment length and RMSD for remote structures reduce to nearly flat line of the match index. This fact indicates clearly that SSM and CE are balanced differently with the respect to alignment length and RMSD, but produce same-quality alignments.


  Figure 4DFR:A-8.
Match Index corresponding to data shown in Figure 4DFR:A-6. Details of the calculations are given here.
 

Z-scores of SSM and CE alignments, shown in Figure 4DFR:A-9 show a relatively good qualitative agreement. SSM Z-scores are more scattered than those of CE. It is interesting to note that CE assignes equal Z-scores to all matches of highly similar structures (PDB entries 10-60), in spite of their difference in RMSD (cf. Figure 4DFR:A-7). These matches have the maximal possible length of Ca-alignments, as may be seen from Figure 4DFR:A-6, therefore one could expect a decrease in Z-score with increasing RMSD. Quantitatively, Z-scores of CE are, on average, twice lower than those of SSM. The latter might be a consequence of different Z-score definitions used by SSM and CE.


  Figure 4DFR:A-9.
Z-scores corresponding to matches shown in Figure 4DFR:A-6. Details of the calculations are given here.
 

 

 

3.  D A L I    (server)

Comparing to VAST and CE, DALI produces less matches (cf. Figure 4DFR:A-10). It fails to align all residues for highly similar structures, including 4DFR:A itself. Ca-alignments from DALI, shown in Figure 4DFR:A-10, allow for the classification to only highly similar and remote structural neighbours, without the intermediate ones, while the latter are clearly present in VAST, CE and SSM results (cf. Figures 4DFR:A-1 and 4DFR:A-6). SSM gives longer (than DALI's) alignments for highly similar structural neighbours and somewhat shorter alignments for the similar ones (seen only in the SSM curve in Figure 4DFR:A-10); the results for remote structural neighbours agree very well.


  Figure 4DFR: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.
 

As Figure 4DFR:A-11 shows, on average SSM produces better (lower) RMSDs of Ca-alignment, than DALI. These RMSDs are achieved at longer alignments (cf. Figure 4DFR:A-10) with the exception for the structures with intermediate level of similarity (PDB entries 61-97).


  Figure 4DFR:A-11.
RMSD of Ca-alignment corresponding to data in Figure 4DFR:A-10. Details of the calculations are given here.
 

A somewhat better quality of SSM alignments is demonstrated by Figure 4DFR:A-12. Indeed, for similar structures, match index of SSM alignments is higher, on average, than that of DALI results. Match indexes for remote structural neighbours agree very well.


  Figure 4DFR:A-12.
Match Index corresponding to data shown in Figure 4DFR:A-10. Details of the calculations are given here.
 

As seen from Figure 4DFR:A-13, DALI assigns Z-scores higher, than SSM's, for highly similar and similar structures, and lower Z-scores for the remote ones. Z-scores from both DALI and SSM clearly show very similar trends, however the quantitative difference is substantial. We found, however, that Z-scores from DALI agree better with minus logarithm of SSM's P-values; as shown by comparisons done on other structures, this finding is of a common nature.


  Figure 4DFR:A-13.
Z-scores corresponding to matches shown in Figure 4DFR:A-10. Details of the calculations are given here.
 

 

 

4.  D E J A V U    (server)

As seen from Figure 4DFR:A-14 below DEJAVU gives just a few (7) highly similar structural neighbours, which is in striking difference with the results obtained from VAST, CE, DALI and SSM. Vast majority of DEJAVU hits relate to remote structural neighbours with Ca-alignments being considerably shorter than those computed by SSM.


  Figure 4DFR: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 4DFR:A-15 reveals that Ca-alignments from DEJAVU have considerably smaller RMSDs than those from SSM, both being in a fairly reasonable range. It therefore may be concluded that DEJAVU and SSM alignments differ in balancing the compromise between RMSD and alignment length.


  Figure 4DFR:A-15.
RMSD of Ca-alignment corresponding to data in Figure 4DFR:A-14. Details of the calculations are given here.
 

The match index, shown in Figure 4DFR:A-16 corroborates the above conclusion. As seen from the Figure, SSM's match indexes are only slightly higher than those from DEJAVU for PDB entries 3-300 (and may be rated as nearly equal for other PDB entries) in spite of considerable differences in alignment lengths and RMSDs (cf. Figures 4DFR:A-14 and 4DFR:A-15).


  Figure 4DFR:A-16.
Match Index corresponding to data shown in Figure 4DFR:A-14. Details of the calculations are given here.
 

P-values, given by SSM and DEJAVU alignments, are shown in Figure 4DFR:A-17. As seen from the Figure, SSM and DEJAVU agree in the assessment of similar structures. At the same time, DEJAVU gives a higher statistical significance of remote structural neighbours.


  Figure 4DFR:A-17.
P-values corresponding to matches shown in Figure 4DFR:A-14. Details of the calculations are given here.
 

Z-scores of SSM and DEJAVU, presented in Figure 4DFR:A-18, show a reasonable agreement for both close and remote structural neighbours.


  Figure 4DFR:A-18.
Z-scores corresponding to matches shown in Figure 4DFR:A-14. Details of the calculations are given here.
 

 

 

5.  Conclusion

The principal quality of 3D Ca-alignments, as measured by match index, is very uniform for all presented results, with maybe some reservations for DALI, which shows a somewhat lower match index for similar structural neighbours. It appears, however, that there is a discrepancy in how different servers balance the compromise between alignment length and RMSD. In that respect, SSM agrees well with VAST, CE and DALI, and shows a noticeable difference of DEJAVU.

The best agreement in Ca-alignment was obtained between VAST and SSM for highly similar and similar structures, and between DALI and SSM for relatively remote structural neighbours. CE tends to longer alignments at higher RMSDs. RMSDs from DALI seem to be higher than those from SSM, at similar alignment lengths. DEJAVU gives shorter alignments and lower RMSDs and fails to find almost all closest structural neighbours.

P-values and Z-scores have similar trends in all cases and thus may be used for rating the results. However, noticeable differences in absolute values of these characteristics are observed between SSM and other servers.