|
Overview
The configuration file for threading is optional, and designed
mainly for advanced users of PROSPECT. If the configuration file is
not available, PROSPECT uses all the default values, which are optimized
for a general purpose. But if a user has some knowledge about the target
protein, the information can be specified in the configuration file and
utilized by PROSPECT.
The usage and format of the .conf files in currently under revision and is
likely to change in upcoming releases of prospect.
Parameter Selections
Weights on Energy Components: The 4 energy components,
i.e., WeightMutate (default 1.0), WeightSingleton (default 0.31), WeightPair
(default 0.31), WeightLoopGap (scaling factor for the loop gap-penalty
function, default 10.4), and WeightGapPenalty (weight for the score of
non-core regions, default 0.91), can be rescaled if a user has evidence
to believe that a certain term is of more importance in a particular case,
e.g.,
WeightMutate 0
WeightSingleton 1.0
WeightPair 0.7
WeightGapPenalty 0.5
MinZscoreCycles (default 100): the minimum number
of cycles of sequence permutations in calculating the Z-score.
MinZscoreCycles 500
MaxLoopSize (default 15): the maximum allowed length
difference between a loop in a structural template and its aligned (loop)
region in the query sequence. e.g.,
MaxLoopSize 20
MaxSkipSeqSize (default 30): the maximum allowed
unaligned residues at the beginning or the end of the template sequence,
or between segments in a structural template (marked by "SEG" in the
template file). If a user wants to do local alignment, or knows that
there might be a large unknown domain insertion between the segments of
the template, a larger number of MaxSkipSeqSize can be specified, e.g.,
MaxSkipSeqSize 200
UsePartitionTree (default 1): a flag to calculate
the pairwise interactions between all the cores (default is on). One
can get a much fast calculation done if using only the pairwise interaction
within cores and between the sequence adjacent cores by setting:
UsePartitionTree 0
ScoreScale (default 1.0): the scaling factor
of the energy function. One may increase it to get a higher accuracy.
In a rare case that the total score is out of the range of "short" type
of the C programming language (-32768 to 32767), "ScoreScale" has to be
decreased, e.g.,
ScoreScale 0.1
Secondary Structure Information
Predicted loop positions: automatically find
loop positions on the target sequence from secondary structure prediction
of PHD, e. g.,
LoopFinder 1
LoopConfidence 9
CoreGrayZone 4
"LoopFinder" (default 0, i.e., off) is the flag whether to use
the feature or not; "LoopConfidence" (default 8) is the minimum reliability
index of loop prediction in a PHD output file; "CoreGrayZone" (default
2) is number of residues at the each end of the cores that allow to be in
the predefined loop regions. Please try the example in "demo/t0068":
thread t0068.seq 1rmg
WeightSStruct (default 0): the flag whether
to use the fitness between the PHD secondary structure prediction and
the secondary structures assigned from a threading alignment as an additional
energy term. The default is off but one can turn it on by:
WeightSStruct 1
ThreadCutOff (default 9999): the upper limit
of the unfavorable total score allowed to output alignments. The default
produces all the alignment outputs. One can view only the alignments with
favorable total scores by setting "ThreadCutOff" 0.0, i.e.,
ThreadCutOff 0.0
|