MI require two maps or a map and a model to calculate scores.
Additionally, map threshold levels can be given for both scores and
for MI the number of layers used to bin the maps can be defined
(default value of 20).
Local scores can be invaluable for assessing the output of
flexible fitting methods such as Flex-EM. Scores such as SMOC
or SCCC can be used to measure the local correlation of the model
with the map. Both SMOC and SCCC are available in the TEMPy
along with example scripts of how to use them. The “score_smoc.py”
example script included in TEMPy takes as input the map file, a
coordinate file and the resolution (see Note 9). The output is given
as a text file of the individual SMOC scores assigned to each residue
(e.g., 336 0.88, where 336 is the residue number and 0.88 the
SMOC score) and can be plotted to a curve using the matplotlib
python library. Additionally, the “get_sccc.py” example script
included in TEMPy can be used to score the SCCC for individual
rigid bodies (see Note 10). The input requires the EM map, PDB
file, resolution, and rigid-body file (that could be produced for
example by RIBFIND). The output is given as a Chimera attribute
file that can be uploaded into Chimera for a 3-D visualization of the
local correlation of rigid bodies with the map.
5.3.3 Ensemble
Generation and Scoring
Different scores in TEMPy can also be used to evaluate an ensemble
of models and identify one or a few best scoring models (generated
by RMSD clustering). A variety of example scripts show workflows
from generating to scoring and clustering of ensembles.
Ensembles of structures can be generated either stochastically
or using angular sweeps. Both methods implemented require an
initial candidate model. The stochastic method requires additional
input of: (a) number of structures to generate; (b) maximum translation and rotation permitted; and (c) graining level for the generation of random vectors (default value of 30). For generating
ensembles using angular sweeps, the user must define (a) axis and
vector for translation; (b) number of structures to generate, (c) and
rotation angle for local rotation. The produced output for both
methods contains a list of generated structures and their
corresponding coordinates.
For scoring of ensemble structures, either a single score or
multiple scores can be used. For the latter, the “Consensus” module can be used (based on Borda counting). The required input is:
(a) a list of the ensemble candidate models (b) an input list of the
scoring functions to use; (c) the target map resolution and sigma
coefficient; and (d) the target map. The output contains the list of
the structures representing different fits, scored and clustered. The
analysis is accompanied by plots and output files that are readable in
Chimera, which can help the user to interpret the consensus among
the scoring metrics chosen [17].
214
Tristan Cragnolini et al.
Additionally, map threshold levels can be given for both scores and
for MI the number of layers used to bin the maps can be defined
(default value of 20).
Local scores can be invaluable for assessing the output of
flexible fitting methods such as Flex-EM. Scores such as SMOC
or SCCC can be used to measure the local correlation of the model
with the map. Both SMOC and SCCC are available in the TEMPy
along with example scripts of how to use them. The “score_smoc.py”
example script included in TEMPy takes as input the map file, a
coordinate file and the resolution (see Note 9). The output is given
as a text file of the individual SMOC scores assigned to each residue
(e.g., 336 0.88, where 336 is the residue number and 0.88 the
SMOC score) and can be plotted to a curve using the matplotlib
python library. Additionally, the “get_sccc.py” example script
included in TEMPy can be used to score the SCCC for individual
rigid bodies (see Note 10). The input requires the EM map, PDB
file, resolution, and rigid-body file (that could be produced for
example by RIBFIND). The output is given as a Chimera attribute
file that can be uploaded into Chimera for a 3-D visualization of the
local correlation of rigid bodies with the map.
5.3.3 Ensemble
Generation and Scoring
Different scores in TEMPy can also be used to evaluate an ensemble
of models and identify one or a few best scoring models (generated
by RMSD clustering). A variety of example scripts show workflows
from generating to scoring and clustering of ensembles.
Ensembles of structures can be generated either stochastically
or using angular sweeps. Both methods implemented require an
initial candidate model. The stochastic method requires additional
input of: (a) number of structures to generate; (b) maximum translation and rotation permitted; and (c) graining level for the generation of random vectors (default value of 30). For generating
ensembles using angular sweeps, the user must define (a) axis and
vector for translation; (b) number of structures to generate, (c) and
rotation angle for local rotation. The produced output for both
methods contains a list of generated structures and their
corresponding coordinates.
For scoring of ensemble structures, either a single score or
multiple scores can be used. For the latter, the “Consensus” module can be used (based on Borda counting). The required input is:
(a) a list of the ensemble candidate models (b) an input list of the
scoring functions to use; (c) the target map resolution and sigma
coefficient; and (d) the target map. The output contains the list of
the structures representing different fits, scored and clustered. The
analysis is accompanied by plots and output files that are readable in
Chimera, which can help the user to interpret the consensus among
the scoring metrics chosen [17].
214
Tristan Cragnolini et al.