COUDES: Supplementary Data

You will find on this page some supplementary information about COUDES:

• list of chains of the different databases
• different set of propensities
• COUDES accuracy on a databse of 426 chains of proteins
• Supplementary figures
• A few references

---

Databases: list of chains

Find here the list of chains for the:

1. database of 426 chains: click here
2. database of 547 chains: click here
3. database of 823 chains: click here

Propensities

Find here the propensities calculated from the:

1. database of 426 chains using:
   • extract_turn: click here
   • PROMOTIF: click here
2. database of 547 chains using:
   • extract_turn: click here
   • PROMOTIF: click here
3. database of 823 chains using:
   • extract_turn: click here
   • PROMOTIF: click here

Coudes accuracy

Here is the global accuracy of COUDES prediction calculated on a seven-fold cross validation on the database PROM426:

	MCC	Qtotal (%)	Qobs (%)	Qpred (%)
turn / non turn	0.42	74.8	69.9	48.8
type I	0.31	84.5	50.0	30.8
type II	0.30	91.0	52.8	22.2
type VIII	0.07	90.7	18.7	6.9
type I'	0.23	94.4	51.8	11.6
type II'	0.11	94.6	32.8	4.6
type IV	0.11	84.9	17.7	20.7

1. MCC = Matthews Correlation Coefficient
2. Qtotal = percentage of correct prediction
3. Qobs = fraction of residues observed in beta-turn that are correctly predicted
4. Qpred = fraction of residues predicted in beta-turn that are correct

For further details, please see Fuchs & Alix, Proteins (2005), in press.

Supplementary figures

You will find here the supplementary figures related to COUDES article [Fuchs 2005] :

• Supplementary Figure 1: ROC curves of COUDES (with or without secondary structure information, and with or without PSSM) with fixed m and PSIthreshold. Each plot has been obtained by varying Sthreshold but with m=4 and PSIthreshold=0 (see Kaur & Raghava (Protein Science, 2003) for a definition of sensitivity and specificity).
  
  
• Supplementary Figure 2: ROC curves of COUDES (with or without secondary structure information, and with or without PSSM) whatever the value of m and PSIthreshold. The red plots were obtained by varying Sthreshold, m and PSIthreshold. The black plots were obtained by varying Sthreshold but with fixed m and PSIthreshold. (m=4 and PSIthreshold=0, like in supplementary figure 1).
  
  

References

• Fuchs PFJ, Alix AJP (2005). High accuracy prediction of beta-turns and their types using propensities and multiple alignments. Proteins, in press.
• Kaur H, Raghava GP (2003). A neural-network based method for prediction of gamma-turns in proteins from multiple sequence alignment. Protein Sci., 12, 627-634.

  ---

  Go to the top

  Go to the query page

  Go to the help page

  ---