About CSI 2.0:

The CSI 2.0 web server is designed to accurately determine protein secondary structures from backbone chemical shifts (Cα, CO, Cβ, N, Hα, NH) and their corresponding protein sequence data. CSI 2.0 accepts either a BMRB (BioMagResBank) formatted (NMR star 2.1 or NMR star 3.1) or a SHIFTY formatted file with at least 70-80% of the backbone atoms having assigned chemical-shifts. The server generates a text-based secondary structure assignment as well as a secondary structure graphics plot in ~90-120 seconds (if the PSIPRED prediction option is not used) or >300 seconds (if the PSIPRED prediction option is used). Extensive testing indicates that CSI 2.0 performs with an accuracy of ~90% (tested on >250 proteins) which is significantly better than other chemical shift based techniques. CSI 2.0 combines the power of machine learning with the added benefits of advanced sequence analysis to significantly increase the reliability of secondary structure identification via chemical shift data. CSI 2.0 also provides a post-prediction Markov-model filtering (7-residue or 5-residue window) to remove the outliers in its final assignment. If you use CSI 2.0 please cite the following reference:

Noor E Hafsa, David S. Wishart: CSI 2.0- a significantly improved version of the Chemical Shift Index, Journal of Biomolecular NMR (2014): Vol. 60(2-3), p. 131-146

If you use backbone RCI result please cite: Mark V. Berjanskii, David S. Wishart (2005) A Simple Method To Predict Protein Flexibility Using Secondary Chemical Shifts.  Journal of the American Chemical Society, 127 (43), 14970 -14971

 Download High-resolution Sequence-Structure Database for Calculating Amino Acid Conformational Preferences


CSI 2.0 Usage

Select the format of input file: NMR-STAR 2.1 (Example) NMR-STAR 3.1 (Example) SHIFTY (Example)

Upload file with chemical shifts here:


Paste chemical shifts into the text box below:
IMPORTANT: If a file is selected for upload above, it will be used and content of the text box below will be ignored.


Additional Options
1) Correct Re-referencing of Chemical Shifts Yes (Default)  No 
2) Correct Neighbor Residue Effects Yes (Default)  No 
3) Use PSIPRED Prediction Yes  (Default)  No 
4) Use Markov Model Post-prediction Filtering 7-res  (Default)  5-res    No 
5) Produce Graphics Output Yes (Default)  No 
6) Predict Flexibility (Random Coil Index) from Chemical Shifts Yes (Default)  No 




Problems? Questions? Suggestions? Contact us!