RAMOS (RApid MOtif Search) / Signature Search / Docking
A three-dimensional signature search of a motif inside a larger cryo-EM
density map can be done using the following SPIDER procedures. Multiple occurrences
of the motif and their locations and orientations inside the searched volume can be found.
A rotationally invariant mask or an asymmetric mask can be used for
calculating locally normalized correlation function. The motif could be
obtained from various sources including crystal structure (if known) or a
single particle reconstruction (if available). Or, alternatively, if a copy of the
searched molecule is recognizable inside the larger volume, it can be cut out
and used as the motif. The searched volume and the motif should have the same
Uses Alan Roseman's Fourier formulation to calculate locally normalized
The SPIDER procedures for doing the signature search are designed to run on a
Linux cluster using PubSub
parallel processing scheduler.
sigs_pub.spi is the main procedure which
to set parameters and another procedure:
sigsloop.spi which does the actual computation.
STEP BY STEP INSTRUCTION FOR RUNNING THE PROGRAM :
For a globular motif, one may use a rotationally invariant mask. The computation will then be much faster than
when using an asymmetric mask. For non-globular motifs, an asymmetric mask should be used.
- Copy sigs_pub.spi,
to the directory where you want to run the program.
- Make sure that SPIDER is callable using the: spider operation.
- Convert the PDB motif into a SPIDER volume by using CP FROM PDB. If the motif is already in SPIDER
format then skip this step. Filter the motif to the same resolution of the large volume if
- Window the motif in x, y and z direction such that it is just contained inside a volume. This
windowed motif is then used as the input volume whose occurrence is being searched inside
the large volume. The windowed motif volume needs not be a cube unless a rotationally invariant
mask is intended to be used.
- Choose a threshold value to create a binary mask of the motif. This mask should resemble the
structure of the motif. Use TH M experimenting with different threshold levels until satisfied.
The resulting threshold value should be given as an input in
- A coarse scanning in all three rotational degrees of freedom should be attempted first.
A reasonable set would be :
PHI = 0 - 350 at 10 degrees interval
THETA = 0 - 180 at 10 degrees interval
PSI = 0 - 350 at 10 degrees interval
These values need to be set in the procedure
sigs_settings.spi. Usually, a coarse scanning
already finds the signatures of the motif at correct x,y,z locations inside the large volume.
- Create a directory "input" using "mkdir ./input". Copy the target volume and the prepared (windowed)
motif into this "input" directory.
- Set all the input parameters in sigs_settings.spi .
- Run sigs_pub.spi using the following command:
Command : ./spider spi/dat @sigs_pub 0 &
- Process the output document files in the "./output" directory using the following Unix commands
and perl programs
(i) cut -c9-100 DOC_OUT_*.dat > DOC_CUT_COMBINED
(Combines the output files. DOC_OUT_* are the ouput files generated
by the signature search procedures. This is done in this way to handle more than a
million keys (lines in the document file) which cann't be handled by SPIDER
(ii) sort -nr -k 7 DOC_CUT_COMBINED > DOC_SORTED
(If any of the cross-correlation coefficient heights is
written in Scientific format in the merged output file, DOC_CUT_COMBINED
then, in LINUX, instead of the above command use :)
(ii) sort -gr -k 7 DOC_CUT_COMBINED > DOC_SORTED
(Sorts the file according to the cross-correlation coefficient heights.)
(i) uniq.perl < DOC_SORTED > DOC_UNIQUE
(Supresses the peaks inside an area equal to the size of the motif around the
highest peak. The procedure continues for the next largest peak after the previous step
is executed. The output file will now contain the unique peaks corresponding
to the probable presence of the signature of the motif at various locations and in different
(ii) make_docfile.perl < DOC_UNIQUE > doc_unique.dat
(Converts the output file from uniq.perl program to SPIDER document file by adding
keys and number of registers etc. info.)
- A finner scan (range is dependent on the angular interval used in coarse scanning) around the
Eulerian angles determined in the coarse scanning step may be done to accurately determine
the position and orientation of the signature of the motif inside the the larger volume (repeat steps 8-10).
- The procedure orient_motif.spi can be used to
create versions of the motif as its signatures are present inside the larger volume. Software EXPLORER can
be used to visualize these versions of the motifs superimposed on the large volume.
- One may use the procedure window.spi for generating an average of the aligned
signatures of the motif (cut and aligned from the large volume) as determined by the program.
- For aiding in visual identification of the locations inside the large volume where the
signatures of the motif are found by the program, the procedures circle.spi and
number.spi may be used.
- If the motif is created using a PDB file, one can rotate and position the pdb co-ordinates
(according to the Euler angles and motif location found by the signature search procedures) using
the SPIDER procedure pdbfit.spi and the PYTHON program
pdbfit.py The rotated PDB volumes can now be visualized by using software "O" or EXPLORER.
- B.K. Rath et al. (2003) Journal of Structural Biology 144, 95-103.
- Alan Roseman (2003) Ultramicroscopy 94, 225-236.
Source file: spider/docs/techs/sigsearch/sigsearch.html