Parameter Reference¶
Transformation¶

genthresh
¶ This parameter defines the cutoff point that will divide the
brain.raw_data
into a set of true signal points and background points based on the probability that each point is true signal. The_Probabilties.h5
dataset is generated after running the Ilastik pixel classification workflow described in Signal Normalization. Pixels with a value of 1 are likely to be background. Correspondingly, pixels with a value close to 0 are most likely to be true signal. We have found that a threshold of 0.5 is sufficient to divide true signal from background; however, if your data contains a lot of intermediate background values (0.40.7), you may benefit from a smaller threshold, e.g. 0.3.Recommended:
0.5

microns
¶ This parameter is a list of the form
[x,y,z]
that specifies the dimensions of the voxel in microns. The data inbrain.raw_data
is scaled bymicrons
to control for datasets in which thez
dimension is larger than thex
andy
dimensions.Example:
[0.16,0.16,0.21]

scale
¶ This is a deprecated parameter that should always be set to
[1,1,1]
.Required:
[1,1,1]

medthresh
¶ This parameter serves the same purpose as
genthresh
; however, it is used exclusively on data used for aligning samples using PCA. This threshold is typically more stringent thangenthresh
to ensure that any noise in the data does not interfere with the alignment process.Recommended:
0.25

radius
¶ The
radius
of the median filter can be tuned to eliminate noisy signal. The typical value forradius
is 20, which refers to the number of neighboring points that are considered in the median filter. A smaller value forradius
will preserve small variation in signal, while a larger value will cause even more blunting and smoothing of the data.Recommended:
20

comporder
¶ This parameter controls how principle components are reassigned to the typical Cartesian coordinate system (XYZ) that most users are familiar with. It takes the form of an array of length 3 that specifies the index of the component that will be assigned to the X, Y, or Z axis:
[x index,y index,z index]
. Please note that the index that matches each principle component starts counting at 0, e.g. 1st PC = 0, 2nd PC = 1, and 3rd PC = 2.For example, if we want to assign the 1st PC to the x axis, the 2nd to the Z axis, and the 3rd to the y axis, the
comporder
parameter would be[0,2,1]
.Example:
[0,2,1]

fitdim
¶ This parameter determines which 2 axes will be used to fit the 2D model. It takes the form of a list of 2 of the 3 dimensions specified as a lowercase string, e.g.
'x','y','z'
.If we wanted to fit a model in the XZ plane, while holding the Y axis constant, the
fitdim
parameter would be['x','z']
.Example:
['x','z']

deg
¶ This parameter specifies the degree of the function that will be fit to the data. The default is
2
, which specifies a parabolic function. A deg of1
would fit a linear function.Default:
2
Warning
The infrastructure to support degrees other than 2 is not currently in place. Check here for updates.
Landmark Calculation¶

anum
¶ This integer specifies the number of divisions along the alpha axis when calculating landmarks. See Selecting anum for guidance on setting this parameter.
Example:
20

tsize
¶ This parameter sets the size of each radial wedge in the landmark calculation. The program works in radians so this parameter should be a float that can evenly divide into 2Pi. We have found that Pi/4 (45º) is a biologically appropriate division for our typical structures.
Example:
0.79

percbins
¶ This parameter is a list of integers that specifies what percentile should be used to calculate the distribution of points along r.
Example:
[50]