The standard curve form controls details of fitting the standard
curve. Any changes made apply only to the selected
experiment (the one whose graphs are shown).
The first item ("Generate standard curves") causes curves to be
fit and the data interpolated. If un-checked, the raw data
in MFI is presented instead. The checkbox can be toggled to
switch between raw and interpolated data.
The second item ("Subtract blank wells...") determines if
background subtraction is turned on. The default is
always off. Background subtraction should be used with
caution; negative data must be discarded on a log scale and cannot
be used for fitting. The only strong justification for
background subtraction is if the negative control probes show high
background levels, indicating non-specific binding of analytes
and/or antibodies to FirePlex particles.
The third item ("Show all curves in plots") controls whether all
the probes or only the selected probes are shown in the individual
and overlay plots. Normally on, the checkbox can be
useful with a 36-plex or 72-plex panel, where the number of plots
becomes overwhelming. When off, the plots are
restricted to whatever subset of probes is selected.
Curve fitting is normally done in log space to give equal weight
to the low end and the high end of the standard curve.
It is possible to fit in linear space with weights to compensate
for the small numerical values at the bottom of the curve; the
usual weight is the inverse square of the concentration. (No
weights are needed for log fitting.)
The fifth item "Calculate RoQ from recovery" defines how the
region of quantitation is defined. Normally it is expected
that all dilutions above the limit of detection are quantitive,
but another commonly used definition is the region where recovered
values are within 25% of nominal values.
For non-immuno assay, e.g. microRNA copy count, the units can be
defined in the next field.
The lower part of the form controls how outliers and detection are defined.
The first item "reject points below detection" causes dilutions
which are below the detection level not to be used in fitting the
Outliers are either those points whose deviation is 3 standard
deviations away from the mean of the other deviations, or those
points which are more than a specified percentage from the
standard curve. If the threshold is set as a number > 1,
it is assumed to mean a number of standard deviations.
If the threshold is to to a number < 1, it is assumed to mean a
fractional deviation. For instance, a value of 3 would mean
outliers are rejected if they are 3 deviations from mean
deviation, while a value of 0.2 would mean outliers are rejected
if they deviate more than 20% from the curve.
Water samples are used to define the MDD. If there are many
water samples (often 6-8 are used) some analytes may show
unusually high values due to accidental contamination. When
there are more than 6 water samples and a measured value has less
than a 1% chance of belonging to the same distribution as the
other water samples, it is considered an outlier. The
outlier test is applied to each analyte separately.
Some organizations used the water value itself as a point on the
curve, with a fictive concentration below the last dilution.
If desired, this type of curve fitting can be turned on by setting
a value of, e.g. 10 in the fictive dilution box, indicating water
samples should be treated as if they are 10X lower in protein
concentration than the last dilution.
Alternatively, the water values can be used to fix the bottom
asymptote of the curve (which corresponds to infinite dilution)
and only the other three parameters are fitted (the top of the
curve, the 50% concentration, and the slope
parameter). While rarely used, this option can be
useful if there are no dilutions close to the lower asymptote.