Structure-based chemical shift predictions

SHIFTS Version 4.1.1, September 2002

Xiao-Ping Xu & David A. Case
Department of Molecular Biology
The Scripps Research Institute

1. About SHIFTS

SHIFTS is a program for predicting 15N, 13Ca, 13Cb, and 13C' chemical shifts from protein structures [8]. It was developed based on an additive model of chemical shift contributions, corresponding to various conformational effects found in a database of density functional theory (DFT) calculations on more than 2000 peptides. Some empirical extensions were used for covering additional conformation regions and residue types. When experimental shifts are available, an optional refinement process for side-chain orientation can also be carried out, which may help identify problems in either the structure or the shift assignments themselves.

Predictions for proton shifts can also be made, using empirical functions [1,2,6]. This is essentially the same model as in earlier versions of SHIFTS. There is also some facility to estimate chemical shift anisotropies, but this is still at a very preliminary stage.

If you have just a few structures to look at, [or if you want (with minimal effort) to see if the results could be of use to you], you can use the Web-based form at the left to ask out server to compute shifts for you.

If you want to run many structures, to see in detail the algorithms, to access different options (such as writing out NMR-Star or Amber format files, etc.), or to modify the codes, you should download the programs to run on your site. This is also the appropriate thing to do if you don't want to send your coordinates across the Internet. Instructions for downloading the source code is at the bottom of this page.

NOTE:You may need to edit your pdb file before submitting it, to remove cofactors or ligands that could confuse our program. Also, many files you get directly from the PDB have missing residues or side chains, and our codes are not always robust in identifying these problems. For example, if a residue is labelled as PHE, but there are no side chain atoms present, our codes don't know how to compute the ring current contribution, and will die. You may need to change some residue names (e.g. to ALA or GLY if the side chain is missing), or use a separate modelling program to fill in the missing pieces.


  1. K. Osapay and D.A. Case. A new analysis of proton chemical shifts in proteins. J. Am. Chem. Soc. 113, 9436-9444 (1991).
  2. K. Ísapay and D.A. Case. Analysis of proton chemical shifts in regular secondary structure of proteins. J. Biomol. NMR 4, 215-230 (1994).
  3. D.A. Case. Calibration of ring current effects in proteins and nucleic acids. J. Biomol. NMR 6, 341-346 (1995).
  4. D. Sitkoff and D.A. Case. Density functional calculations of proton chemical shifts in model peptide systems. J. Am. Chem. Soc. 119, 12262-12273 (1997).
  5. D. Sitkoff and D.A. Case. Theories of chemical shift anisotropies in proteins and nucleic acids. Prog. NMR Spectr. 32, 165-190 (1998).
  6. A.P. Dejaegere, R.A. Bryce and D.A. Case. An empirical analysis of proton chemical shifts in nucleic acids. In Modeling NMR Chemical Shifts, J.C. Facelli and A.C. de Dios, eds. (Washington, American Chemical Society, 1999), pp. 194-206.
  7. D.S. Wishart and D.A. Case. Use of chemical shifts in macromolecular structure determination. Meth. Enzymol. 338, 3-34 (2001).
  8. X.P. Xu and D.A. Case. Automated prediction of 15N, 13Ca, 13Cb and 13C' chemical shifts in proteins using a density functional database. J. Biomol. NMR 21 321-333 (2001).
  9. X.P. Xu and D.A. Case. Probing multiple effects on 15N, 13Ca, 13Cb and 13C' chemical shifts in peptides using density functional theory. Biopolymers 65 408-423 (2002).

2. Preparation of PDB files

SHIFTS tries to read in PDB files as given by Brookhaven standards, but is not always successful. Here are some things to look for in preparing your PDB file:

3. Preparation of OBS files

If you have experimental shifts data available and would like to see a comparison between the predicted and observed shifts, you may upload a file with the observed shifts. There are two formats we support. The first (which must have a ".obs" extension) is a simple heavy-atom shift file that looks like the following:

# 1ACF.obs
# resNo.  res       N        Ca       Cb        C'
   3       Q     118.26    58.53    28.48      0.00
   4       T     114.00    66.23    68.58      0.00
   5       Y     119.55    60.68    38.29    177.50
   6       V     115.86	   66.72    32.26    177.10
   7       D     118.29    57.94    40.88    178.90
   8       T     115.72    65.55    68.82    175.00
   9       N     116.09    56.61    39.01    174.20
  10       L     114.26    55.50    41.90    176.70

Please be sure if the residue numbers in your obs file are same as that in your pdb file. The optional observed-shifts-file is used to compare calculated and observed shifts as well as do side-chain orientation refinement.

You may also use a BMRB star file (with an "str" extension). This format file is a subset of the data in an NMR -star file from BMRB. An example is:

  # your comments
   1   ALA  C     C  172.00   .  1
   1   ALA  CA    C   51.00   .  1
   1   ALA  HA    H    4.40   .  1
   1   ALA  CB    C   17.90   .  1
   1   ALA  HB    H    1.57   .  1
   2   PRO  C     C  176.80   .  1
   2   PRO  CA    C   63.10   .  1
   2   PRO  HA    H    4.62   .  1
   2   PRO  CB    C   31.90   .  1
   2   PRO  HB2   H    2.39   .  2

4. Output abbreviations

Abbreviations for heavy atom shift contributions:
    bb-p ...... preceding backbone effect
    bb-s ...... self backbone effect
    bb-f ...... following backbone effect
    chi-p ..... preceding chi effect
    chi-s ..... self chi effect
    HB-D....... direct HB effect (NH <-)
    HB-I....... indirect HB effect (C=O <)
    REF ....... reference shifts for this amino acid
    pred....... predicted shift = sum of the above
    obs ....... observed shift (if provided)

    Lines marked "diff" indicate some difference in sequence between the pdb
    file and the obs file.

Abbreviations for the proton shift contributions:
    RingCur.... ring currents from aromatic rings
    El ........ electrostatic contribution (backbone only for now)
    P_anis .... peptide group anisotropy
    Const ..... constant constribution
    RC ........ random coil shift
    pred ...... predicted shift = sum of the above

    Note that the message "unmatched atom ...OXT" can be ignored.  In the
    ouptut, shifts for methyl protons are reported as the average for the
    three protons, and the atom name of the last proton is used; for
    example, the (averaged) methyl shift for alanine would be labelled

5. Sample inputs and outputs

6. Contact us

This web-based service is new (begun April, 2002), and there may well be be glitches and/or more serious errors. If you have suggestions or questions, please send email to:

7. Download the sources.

The SHIFTS code can be obtained from:


You will need the NAB program (version 4.3.2 or later, and also available at the above Web site) in order to compile SHIFTS. Both NAB and SHIFTS are distributed under the GNU General Public License (GPL).