Skrmdb Manual
User Manual:
Open the PDF directly: View PDF 
.
Page Count: 11
Package ‘skrmdb’
August 3, 2018
Type Package
Title Package to estimate ED50
Version 4.2.5
Date 2018-08-03
Author David Siev
Maintainer Marie Vendettuoli <marie.c.vendettuoli@aphis.usda.gov>
Description Package to estimate ED50 by the methods of Spearman-Karber,
Reed-Muench, Dragstedt-Behrens. No endorsement, claim, or warranty is
implied for this package. It is made available for investigational or pedagogical use only
License MIT + file LICENSE
LazyLoad yes
Depends R (>= 3.4.4)
Imports methods, stats
Collate 'class-skrmdb.r' 'skrmdb-package.r' 'aaa.r' 'bbb.r'
'dragbehr.r' 'reedmuench.r' 'skrmdb.R'
RoxygenNote 6.0.1
NeedsCompilation no
Rtopics documented:
skrmdb-package....................................... 2
DragBehr .......................................... 2
getED ............................................ 4
ReedMuench ........................................ 5
sk-class ........................................... 7
SKRMDB-class....................................... 7
SpearKarb.......................................... 8
Index 11
1
2DragBehr
skrmdb-package skrmdb package.
Description
Includes functions to estimate ED50 by the methods of Spearman-Karber, Reed-Muench, Dragstedt-
Behrens.
For internal use only at the USDA Center for Veterinary Biologics
Details
Package: skrmdb-package
Type: Package
Version: 4.2.5
Date: 2018-08-03
License: MIT
LazyLoad: yes
LazyData: yes
Resources
• QUICK START: https://github.com/ABS-dev/skrmdb/blob/master/README.md
• PACKAGE VIGNETTE: https://www.aphis.usda.gov/animal_health/vet_biologics/
publications/STATWI0001.pdf
• BUG REPORTS: https://github.com/ABS-dev/skrmdb/issues
Author(s)
David Siev <David.Siev@aphis.usda.gov>
DragBehr Dragstedt-Behrens estimator
Description
Gives the Dragstedt-Behrens estimate of median effective dose (ED50)
Usage
DragBehr(formula = NULL, data = NULL, y, n, x, warn.me = T, show = F)
DragBehr 3
Arguments
formula a formula of the form cbind(y,n) ~ x
data a data frame
ythe number responding (response should be monotone increasing)
nthe group size
xlog dilution or dose
warn.me boolean to give warning message
show boolean to display extended summary
Details
Data input may either be a formula and data frame, or variables may be input directly (see example).
The Dragstedt-Behrens method estimates the median effective dose by interpolating between the
two doses that bracket the dose producing median response. It accumulates sums in both directions
by assuming that those that responded at a lower dose would respond at a higher dose, and those
that did not respond at a higher dose would not respond at a lower dose. The hypothetical fraction
that would have responded at a particular dose is estimated from the cumulative sums at that dose.
The ED50 is estimated by interpolation on the line that connects the hypothetical fractions of the
bracketing doses.
Value
object of SKRMDB class with slots:
ed Estimated median effective dose (ED50)
eval Evaluation method: ’DragBehr’
Note
Many microbiology texts mistakenly present the Dragstedt-Behrens method as the Reed-Muench
method.
And yes, it is absurd to have an R function for an archaic method developed to avoid complex cal-
culations.
Input data is expected to be sorted by x (either increasing or decreasing). Use of unsorted data will
result in error.
Author(s)
David Siev <David.Siev@aphis.usda.gov>
References
Miller, Rupert G. (1973). Nonparametric estimateors of the mean tolerance in bioassay. Biometrika.
60: 535 - 542.
Behrens, B. (1929) Zur Auswertung der Digitalisblatter im Froschversuch. Arkiv fur Experimentelle
4getED
Pathologie und Pharmakologie. 140: 237-256.
Dragstedt, CA., Lang, VF. (1928). Respiratory Stimulants in acute poisoning in rabbits. J. of
Pharmacology 32: 215–222.
See Also
The function ReedMuench gives the Reed-Muench estimate of ED50, SKRMDB-class
Examples
X <- data.frame(dead=c(0,3,5,8,10,10),total=rep(10,6),dil=1:6)
DragBehr(cbind(dead,total) ~ dil, X)
# or
DragBehr(y=c(0,3,5,8,10,10), n=rep(10,6), x=1:6)
# db
# 2.906593
## Not run:
## unordered data
X2 <- data.frame(dead = c(10,8,5,3,0), total = rep(10, 5), dil = c(1, 3, 2, 4, 5))
DragBehr(cbind(dead,total) ~ dil, X2)
DragBehr(y = X2$dead, n = X2$total, x = X2$dil)
## monotone decreasing (note that x variable direction is ignored!!)
reverse <- data.frame(dead = c(10, 8, 5, 3, 0), total = rep(10, 5), dil = 5:1)
DragBehr(cbind(dead,total) ~ dil, reverse)
DragBehr(y = reverse$dead, n = reverse$total, x = reverse$dil)
## not monotone
X3 <- data.frame(dead = c(1:3, 5, 4), total = rep(10, 5), dil = 1:5 )
DragBehr(cbind(dead, total) ~ dil, X3)
DragBehr(y = X3$dead, n = X3$total, x = X3$dil)
## End(Not run)
getED Accessor to retrieve the numeric median effective dose from a SKR-
MDB or sk object
Description
This is an accessor function for retrieving the numeric value of effective from a SKRMDB or sk
data object object generated by ReedMuench, SpearKarb or DragBehr
Usage
getED(x)
## S4 method for signature 'SKRMDB'
getED(object)
ReedMuench 5
Arguments
object class SKRMDB or sk
Author(s)
Marie Vendettuol <marie.c.vendettuoli@aphis.usda.gov>
See Also
SKRMDB-class,sk-class
Examples
## with an object of class SKRMDB
temp1 <- DragBehr(y=c(0,3,5,8,10,10), n=rep(10,6), x=1:6)
getED(temp1)
## with an object of class sk
X <- data.frame(dead=c(0,3,5,8,10), total=rep(10,5), dil=1:5)
temp2 <- SpearKarb(cbind(dead,total) ~ dil, X)
getED(temp2)
ReedMuench Reed-Muench estimator
Description
Gives the Reed-Muench estimate of median effective dose (ED50)
Usage
ReedMuench(formula = NULL, data = NULL, y, n, x, warn.me = T, show = F)
Arguments
formula a formula of the form cbind(y,n) ~ x
data a data frame
ythe number responding (response should be monotone increasing)
nthe group size
xlog dilution or dose
warn.me boolean to give warning message
show boolean to display extended summary
Details
Data input may either be a formula and data frame, or variables may be input directly (see example).
The Reed-Muench method estimates the median effective dose by interpolating between the two
doses that bracket the dose producing median response. It accumulates sums in both directions that
represent the hypothetical number that would have responded or not at each dose. It does so by
assuming that those that responded at a lower dose would respond at a higher dose, and those that
did not respond at a higher dose would not respond at a lower dose. The ED50 is the intersection of
the lines connecting the two sets of cumulative sums between the bracketing doses.
6ReedMuench
Value
object of class SKRMDB with slots:
ed Estimated median effective dose (ED50)
eval Evaluation method: ’ReedMuench’
Note
Many microbiology texts mistakenly present the Dragstedt-Behrens method as the Reed-Muench
method.
And yes, it is absurd to have an R function for an archaic method developed to avoid complex cal-
culations.
Input data is expected to be sorted by X variable (either increasing or decreasing). Use of unsorted
X variables will result in error. Y variables are evaluated for monotone, increasing or decreasing;
however the estimate will be calculated in the original order regardless.
Author(s)
David Siev <David.Siev@aphis.usda.gov>
References
Miller, Rupert G. (1973). Nonparametric estimateors of the mean tolerance in bioassay. Biometrika.
60: 535 – 542.
Reed LJ, Muench H (1938). A simple method of estimating fifty percent endpoints. American
Journal of Hygiene. 27:493–497.
See Also
The function DragBehr gives the Dragstedt-Behrens estimate of ED50 SKRMDB-class
Examples
X <- data.frame(dead=c(0,3,5,8,10,10),total=rep(10,6),dil=1:6)
ReedMuench(cbind(dead,total) ~ dil, X)
# or
ReedMuench(y=c(0,3,5,8,10,10), n=rep(10,6), x=1:6)
# rm
# 2.916667
## Not run:
## unordered data
X2 <- data.frame(dead = c(10,8,5,3,0), total = rep(10, 5), dil = c(1, 3, 2, 4, 5))
ReedMuench(cbind(dead,total) ~ dil, X2)
ReedMuench(y = X2$dead, n = X2$total, x = X2$dil)
sk-class 7
## monotone decreasing (note that x variable direction is ignored!!)
reverse <- data.frame(dead = c(10, 8, 5, 3, 0), total = rep(10, 5), dil = 5:1)
ReedMuench(cbind(dead,total) ~ dil, reverse)
ReedMuench(y = reverse$dead, n = reverse$total, x = reverse$dil)
## not monotone
X3 <- data.frame(dead = c(1:3, 5, 4), total = rep(10, 5), dil = 1:5 )
ReedMuench(cbind(dead, total) ~ dil, X3)
ReedMuench(y = X3$dead, n = X3$total, x = X3$dil)
## End(Not run)
sk-class Class definition for sk object
Description
The sk object holds values for the Spear Karb estimator for median estimated dose. It extends the
SKRMDB data object with value for variance.
Details
eval Evaluation method. "SpearKarb". Character string.
ed Median effective dose by eval method. Numeric.
sk.var variance. Numeric.
Author(s)
Marie Vendettuoli <marie.c.vendettuoli@aphis.usda.gov>
See Also
SKRMDB-class
Examples
new('sk', sk.var = 0.06888889, ed = 2.9, eval = "SpearKarb")
SKRMDB-class Class definition for SKRMDB object
Description
The SKRMDB object holds output from functions in the SKRMDB package.
Details
eval Evaluation method. One of ’ReedMeunch’, ’SpearKarb’, or ’DragBehr’. Character string.
ed Median effective dose by eval method. Numeric.
8SpearKarb
Author(s)
Marie Vendettuoli <marie.c.vendettuoli@aphis.usda.gov>
See Also
sk-class
Examples
new("SKRMDB", ed = 2.906593, eval = "DragBehr")
SpearKarb Spearman-Karber estimator
Description
Gives the Spearman-Karber estimate of the mean effective dose
Usage
SpearKarb(formula = NULL, data = NULL, y, n, x)
Arguments
formula a formula of the form cbind(y,n) ~ x
data a data frame
ythe number responding (response should be monotone increasing)
nthe group size
xlog dilution or dose
Details
Data input may either be a formula and data frame, or variables may be input directly (see example).
The Spearman-Karber method gives a non-parametric estimate of the mean of an tolerance dis-
tribution from its empirical distribution (EDF). The empirical PMF is derived from the EDF by
differencing and the estimator is Pxf (x). If the EDF does not cover the entire support of x,
SpearKarb() extends it by assuming the next lower dilution would produce zero response and the
next higher dilution would produce complete response.
Value
object of class sk
ed estimator of mean response
sk.var variance
eval evaluation method: ’SpearKarb’
SpearKarb 9
Note
Input data is expected to be sorted by X variable (either increasing or decreasing). Use of unsorted
X variables will result in error. Y variables are evaluated for monotone, increasing or decreasing;
however the estimate will be calculated in the original order regardless of direction.
Author(s)
David Siev <David.Siev@aphis.usda.gov>
References
Miller, Rupert G. (1973). Nonparametric estimateors of the mean tolerance in bioassay. Biometrika.
60: 535 – 542.
Karber, G. (1931). Beitrag zur kollektiven Behandlung Parmakogischer Reihenversuche. Archiv
fur Experimentelle Pathologie und Pharmakologie. 162: 480–487.
Spearman, C. (1908). The method of "right and wrong cases" ("constant stimuli") without Gauss’s
formulae. Brit J. of Psychology. 2: 227–242.
Examples
X <- data.frame(dead=c(0,3,5,8,10), total=rep(10,5), dil=1:5)
SpearKarb(cbind(dead,total) ~ dil, X)
# sk sk.var
#2.90000000 0.06888889
# without zero and complete response
X <- data.frame(dead=c(3,5,8),total=rep(10,3),dil=2:4)
SpearKarb(cbind(dead,total) ~ dil, X)
# or
SpearKarb(y=c(3,5,8), n=rep(10,3), x=2:4)
# sk sk.var
#2.90000000 0.06888889
## Not run:
## unordered
X2 <- data.frame(dead = c(10,8,5,3,0), total = rep(10, 5), dil = c(1, 3, 2, 4, 5))
SpearKarb(cbind(dead,total) ~ dil, X2)
SpearKarb(y = X2$dead, n = X2$total, x = X2$dil)
## monotone decreasing (note that x variable direction is ignored!!)
reverse <- data.frame(dead = c(10, 8, 5, 3, 0), total = rep(10, 5), dil = 5:1)
SpearKarb(cbind(dead,total) ~ dil, reverse)
SpearKarb(y = reverse$dead, n = reverse$total, x = reverse$dil)
## not monotone
X3 <- data.frame(dead = c(1:3, 5, 4), total = rep(10, 5), dil = 1:5 )
SpearKarb(cbind(dead, total) ~ dil, X3)
SpearKarb(y = X3$dead, n = X3$total, x = X3$dil)
10 SpearKarb
## End(Not run)
