Dear RESCUE2 user,
You have been using the RESCUE2 Web service, and you would like now to use
the program at home. We have been designing RESCUE2 for using it mainly
through a Web server, so running it at home can be slightly more complicated.
Here are some advices which can help you to do so.
1) You need absolutely the file BMRB_stats.out, which contained the
statistics on the chemical shifts (mean values and standard deviations),
calculated from a list BMRB star files. The file BMRB_stats.out can be
downloaded from: https://bioserv.cbs.cnrs.fr/rescue2/download.html, as well as
the list of BMRB STAR files (BMRB_selected_sequences.txt). These STAR files
correspond to a BMRB state in October 2003. You may want to update the
statistics using more recent versions of BMRB, and the archive
BMRB_tools-1.0.tar.gz contain all tools to do it, but they are not very
documented. If you need help, ask me, I could try to ansver questions about
these tools.
2) The program rescue2 is in the archive RESCUE2-1.0.tar.gz. To compile it,
you need to install the gsl library 3.0. The program was not tested with other
library versions. The gsl library is mandatory, as it calculates the
permutations and combinations of chemical shift values, to handle different
assignment possibilities (see reference: Marin A, Malliavin TE, Nicolas P,
Delsuc MA. From NMR chemical shifts to amino acid types: investigation of
the predictive power carried by nuclei. J Biomol NMR. 2004 Sep;30(1):47-60
for details).
3) The syntax to run RESCUE2 is the following:
./rescue2 -cspi -sf BMRB_stats.out -af cs.in < bmr.xml > bmr.pred
The option -cspi is telling you that you are calculating selected permutation
independant probabilities, the -sf option is giving you the name of the
statistics file, and the -af option the name of the file containing atoms to
analyse, in order to know what are all the possible choices for the
probabilities calculation. For example, if you are sure to not have a type
of nuclei or a type of amino-acid in your data, you have to remove it from
the cs.in. You can remove nuclei and/or amino-acids by typing a '#' to comment
the corresponding lines. By default, the cs.in file given in the archives
contains all the possible nuclei for all the amino-acids, and you can use it
as a first approximation. All the possible options are explained by typing
'rescue2 -h'.
The xml input format of the input chemical shifts in xml format (bmr.xml) is:
4.09 2.01 1.96 2.55 2.53 2.09
53.8 35.1 31.1 16.9
4.21 2.33 1.91 1.92 1.86 3.23 3.23
62.5 32.8 27.1 49.5
8.11 4.01 1.92 1.87 1.72 1.71 3.23 3.23
118.1
178.5 59.6 29.8 27.2 43.3
One important remark is that RESCUE2 is very pedantic with its interpretation
of the XML input file. In particular, you have to respect strictly
the indentation of the lines, with two blanks of indentation in tge
and lines, and 4 blanks in the chemical shift lines.
Each spin system is contained between the tags and .
The type=" " is to give the real amino-acid type when you know it for testing
purposes. So, for prediction, you have to let it empty. Inside each spin
system, you can list the chemical shifts (tags ) sorted by the
nuclei type (N, C or H). If one nucleus is not present in the data, you
are just removing the corresponding ... . For each set of chemical
shifts, you need to precise the nucleus type (H, C or N) and the number of
chemical shifts.
There is a preprocessing perl script (prep_rescue2.pl, available on the page:
https://bioserv.cbs.cnrs.fr/rescue2/download.html) for preparing the xml input
for RESCUE2. This perl script is relatively rough.
The results you are getting with the previous input are:
1: M (1.00)
2: P (1.00)
3: R (0.97) K (0.03)
For each of the spin systems, you have the most probable amino-acid types,
with the probability value in parentheses.
Thank you to Woo-Sung Son from the National Research Laboratory for Membrane
Protein Structure, Seoul National University, for having pushed me to write
this tutorial
I hope that these remarks will be useful for you. Please let me know whether
something is unclear in it, and I shall try to improve it.
Good luck with RESCUE2 !
Thérèse Malliavin