aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorJohannes Ranke <jranke@uni-bremen.de>2017-03-24 15:28:44 +0100
committerJohannes Ranke <jranke@uni-bremen.de>2017-03-24 15:28:44 +0100
commitbaf8e4341b58c09532b4ca37ebf80c9068934db4 (patch)
treea6abd8cbdfa849a672d2abcfde5af4bfc180bcf6
parent22c11d5dbc118881f9cfd7c9c2ac9711295a0f16 (diff)
Import version 0.6.7 from CRAN
I had deleted the git commits between 0.6.4 and 0.6.7 while removing drfit from github, while the latest changes were not pushed to jrwb.de yet.
-rw-r--r--ChangeLog8
-rw-r--r--DESCRIPTION24
-rw-r--r--NAMESPACE3
-rw-r--r--R/checkexperiment.R28
-rw-r--r--R/checksubstance.R52
-rw-r--r--R/drcfit.R68
-rw-r--r--R/drdata.R2
-rw-r--r--R/drfit.R47
-rw-r--r--R/drplot.R36
-rw-r--r--data/XY.rdabin1602 -> 630 bytes
-rw-r--r--inst/doc/drfit-Rnews.pdfbin229223 -> 169560 bytes
11 files changed, 140 insertions, 128 deletions
diff --git a/ChangeLog b/ChangeLog
index dce0368..170eab9 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,11 @@
+2016-09-02 Johannes Ranke
+
+ * Make drcfit compatible with drc >= 3.0.1 (the row names of the matrix
+ returned by the ED() function were changed).
+ * Switch back to using requireNamespace(RODBC) in order to address
+ a NOTE issued by R CMD check
+ * Use single quotes to quote other packages in the DESCRIPTION file
+
2015-12-15 Johannes Ranke
* Use require(RODBC) instead of requireNamespace(RODBC) as the former fails
diff --git a/DESCRIPTION b/DESCRIPTION
index 173e9c6..4a0f79b 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -1,10 +1,10 @@
Package: drfit
-Version: 0.7.1
-Date: 2017-03-24
+Version: 0.6.7
+Date: 2016-09-02
Title: Dose-Response Data Evaluation
Authors@R: c(person("Johannes", "Ranke", role = c("aut", "cre"),
email = "jranke@uni-bremen.de"))
-Imports: graphics, grDevices, MASS, drc, reshape2, qcc
+Imports: graphics, grDevices, MASS, drc
Suggests: RODBC
Description: A somewhat outdated package of basic and easy-to-use functions for
fitting dose-response curves to continuous dose-response data, calculating some
@@ -12,16 +12,22 @@ Description: A somewhat outdated package of basic and easy-to-use functions for
the more powerful and actively developed 'drc' package. Functions that are
fitted are the cumulative density function of the lognormal distribution
(probit fit), of the logistic distribution (logit fit), of the weibull
- distribution (weibull fit) and a linear-logistic model ("linlogit" fit),
+ distribution (weibull fit) and a linear-logistic model ('linlogit' fit),
derived from the latter, which is used to describe data showing stimulation at
low doses (hormesis). In addition, functions checking, plotting and retrieving
- dose-response data retrieved from a database accessed via RODBC are included.
- As an alternative to the original fitting methods, the algorithms from the drc
+ dose-response data retrieved from a database accessed via 'RODBC' are included.
+ As an alternative to the original fitting methods, the algorithms from the 'drc'
package can be used.
Encoding: latin1
License: GPL (>= 2)
LazyLoad: yes
LazyData: yes
-URL: http://www.r-project.org,
- http://www.uft.uni-bremen.de/chemie/ranke?page=drfit,
- http://cgit.jrwb.de/drfit/
+URL: http://www.r-project.org,
+ http://www.uft.uni-bremen.de/chemie/ranke?page=drfit,
+ http://cgit.jrwb.de/drfit/
+NeedsCompilation: no
+Packaged: 2016-09-02 14:24:52 UTC; jranke
+Author: Johannes Ranke [aut, cre]
+Maintainer: Johannes Ranke <jranke@uni-bremen.de>
+Repository: CRAN
+Date/Publication: 2016-09-02 21:28:14
diff --git a/NAMESPACE b/NAMESPACE
index 120d052..cadcc62 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -8,5 +8,4 @@ import(
)
importFrom("drc", drm, ED, LN.2, LL.2, BC.4, W1.2)
-importFrom("reshape2", melt, acast)
-importFrom("qcc", qcc)
+importFrom("utils", "packageVersion")
diff --git a/R/checkexperiment.R b/R/checkexperiment.R
index b69b81f..8c2f472 100644
--- a/R/checkexperiment.R
+++ b/R/checkexperiment.R
@@ -25,21 +25,21 @@ checkexperiment <- function(id, db = "ecotox", endpoint = "%")
" WHERE ", exptype, " = ", id)
commentdata <- RODBC::sqlQuery(channel,commentquery)
comment <- as.character(commentdata[[1]])
-
+
expquery <- paste("SELECT experimentator,substance, ",
testtype, ",conc,unit,", responsename, ",performed,ok",
- " FROM ",db," WHERE ",exptype,"=", id,
+ " FROM ",db," WHERE ",exptype,"=", id,
sep = "")
if (db == "ecotox") {
- expquery <- paste(expquery, " AND type LIKE '",
+ expquery <- paste(expquery, " AND type LIKE '",
endpoint, "'", sep = "")
}
expdata <- RODBC::sqlQuery(channel,expquery)
if (db %in% c("cytotox","enzymes")) {
- controlquery <- paste("SELECT type,response FROM controls
+ controlquery <- paste("SELECT type,response FROM controls
WHERE plate=",id)
controldata <- RODBC::sqlQuery(channel,controlquery)
}
@@ -63,7 +63,7 @@ checkexperiment <- function(id, db = "ecotox", endpoint = "%")
numberOfBlinds <- NA
meanOfBlinds <- NA
stdOfBlinds <- NA
-
+
}
numberOfControls <- length(controls$response)
if (numberOfControls > 0) {
@@ -78,7 +78,7 @@ checkexperiment <- function(id, db = "ecotox", endpoint = "%")
if (length(expdata$experimentator) < 1) {
stop("There is no response data for ",exptype," ",
id," in database ",db,"\n")
- }
+ }
exptypestring <- paste(toupper(substring(exptype,1,1)),
substring(exptype,2),sep="")
expdata$experimentator <- factor(expdata$experimentator)
@@ -87,7 +87,7 @@ checkexperiment <- function(id, db = "ecotox", endpoint = "%")
expdata$substance <- factor(expdata$substance)
expdata$unit <- factor(expdata$unit)
expdata$ok <- factor(expdata$ok)
-
+
cat("\n",exptypestring,id,"from database",db,":\n\n",
"\tExperimentator(s):\t",levels(expdata$experimentator),"\n",
"\tType(s):\t\t",levels(expdata$type),"\n",
@@ -100,7 +100,7 @@ checkexperiment <- function(id, db = "ecotox", endpoint = "%")
"\tblind\t",numberOfBlinds,"\t",meanOfBlinds,"\t",stdOfBlinds,"\n",
"\tcontrol\t",numberOfControls,"\t",meanOfControls,"\t",
stdOfControls,"\t\t",percentstdOfcontrols,"\n")
-
+
if (db == "ecotox") {
boxplot(controls$response,
@@ -116,22 +116,22 @@ checkexperiment <- function(id, db = "ecotox", endpoint = "%")
boxwex=0.4,
main=paste("Plate ",id))
}
-
+
drdata <- expdata[c(2,4,6)]
drdata$substance <- factor(drdata$substance)
substances <- levels(drdata$substance)
-
+
lld <- log10(min(subset(drdata,conc!=0)$conc))
lhd <- log10(max(drdata$conc))
plot(1,type="n",
- xlim=c(lld - 0.5, lhd + 2),
- ylim= c(-0.1, 2),
+ xlim=c(lld - 0.5, lhd + 2),
+ ylim= c(-0.1, 2),
xlab=paste("decadic logarithm of the concentration in ",levels(expdata$unit)),
ylab=responsename)
-
+
drdatalist <- split(drdata,drdata$substance)
-
+
for (i in 1:length(drdatalist)) {
points(log10(drdatalist[[i]]$conc),drdatalist[[i]][[responsename]],col=i);
}
diff --git a/R/checksubstance.R b/R/checksubstance.R
index 06cc635..3e07f92 100644
--- a/R/checksubstance.R
+++ b/R/checksubstance.R
@@ -1,7 +1,7 @@
checksubstance <- function(substance, db = "cytotox", experimentator = "%",
- celltype = "%", enzymetype = "%", organism = "%",
+ celltype = "%", enzymetype = "%", organism = "%",
endpoint = "%",
- whereClause = "1", ok= "%")
+ whereClause = "1", ok= "%")
{
databases <- data.frame(
responsename=c("viability","activity","response"),
@@ -23,13 +23,13 @@ checksubstance <- function(substance, db = "cytotox", experimentator = "%",
if (db == "cytotox") {
type <- celltype
- }
+ }
if (db == "enzymes") {
type <- enzymetype
- }
+ }
if (db == "ecotox") {
type <- organism
- }
+ }
query <- paste("SELECT experimentator,substance,",
testtype, ",", exptype, ",conc,unit,",responsename,",ok",
@@ -41,7 +41,7 @@ checksubstance <- function(substance, db = "cytotox", experimentator = "%",
sep = "")
if (db == "ecotox") {
- query <- paste(query, " AND type LIKE '",
+ query <- paste(query, " AND type LIKE '",
endpoint, "'", sep = "")
}
@@ -51,39 +51,39 @@ checksubstance <- function(substance, db = "cytotox", experimentator = "%",
if (length(data$experimentator) < 1) {
stop(paste("\nNo response data for",substance,"in database",
db,"found with these parameters\n"))
- }
-
- data$experimentator <- factor(data$experimentator)
+ }
+
+ data$experimentator <- factor(data$experimentator)
data$substance <- factor(data$substance)
- substances <- levels(data$substance)
- data$type <- factor(data[[testtype]])
- data[[exptype]] <- factor(data[[exptype]])
+ substances <- levels(data$substance)
+ data$type <- factor(data[[testtype]])
+ data[[exptype]] <- factor(data[[exptype]])
experiments <- levels(data[[exptype]])
concentrations <- split(data$conc,data$conc)
concentrations <- as.numeric(names(concentrations))
- data$unit <- factor(data$unit)
+ data$unit <- factor(data$unit)
data$ok <- factor(data$ok)
if (length(experiments)>6) {
- palette(rainbow(length(experiments)))
+ palette(rainbow(length(experiments)))
}
-
+
plot(log10(data$conc),data[[responsename]],
- xlim=c(-2.5, 4.5),
- ylim= c(-0.1, 2),
- xlab=paste("decadic logarithm of the concentration in ",levels(data$unit)),
- ylab=responsename)
-
+ xlim=c(-2.5, 4.5),
+ ylim= c(-0.1, 2),
+ xlab=paste("decadic logarithm of the concentration in ",levels(data$unit)),
+ ylab=responsename)
+
explist <- split(data,data[[exptype]])
-
- for (i in 1:length(explist)) {
- points(log10(explist[[i]]$conc),explist[[i]][[responsename]],col=i);
- }
-
+
+ for (i in 1:length(explist)) {
+ points(log10(explist[[i]]$conc),explist[[i]][[responsename]],col=i);
+ }
+
legend("topleft", experiments, pch=1, col=1:length(experiments), inset=0.05)
title(main=paste(substance," - ",levels(data$experimentator)," - ",levels(data$type)))
- exptypename <- paste(toupper(substring(exptype,1,1)),
+ exptypename <- paste(toupper(substring(exptype,1,1)),
substring(exptype,2), sep = "")
experimentators <- paste(levels(data$experimentator), collapse = " ")
types <- paste(levels(data$type), collapse = " ")
diff --git a/R/drcfit.R b/R/drcfit.R
index 0d27c1c..008c53c 100644
--- a/R/drcfit.R
+++ b/R/drcfit.R
@@ -14,7 +14,7 @@ drcfit <- function(data, chooseone=TRUE,
# model result was appended
rsubstance <- array() # the substance names in the results
rndl <- vector() # number of dose levels
- rn <- vector() # mean number of replicates
+ rn <- vector() # mean number of replicates
# in each dose level
runit <- vector() # vector of units for each result row
rlhd <- rlld <- vector() # highest and lowest doses tested
@@ -27,6 +27,16 @@ drcfit <- function(data, chooseone=TRUE,
models <- list() # a list containing the dose-response models
splitted <- split(data,data$substance)
+
+ # The indexing of the results of the ED50 function changed with drc 3.0.1
+ if (packageVersion("drc") > 3) {
+ ED50_row_index = "e:1:50"
+ EDx_row_index_prefix = "e:1"
+ } else {
+ ED50_row_index = "1:50"
+ EDx_row_index_prefix = "1"
+ }
+
for (i in substances) {
tmp <- splitted[[i]]
fit <- FALSE
@@ -69,8 +79,7 @@ drcfit <- function(data, chooseone=TRUE,
active <- FALSE
if (linlogit)
{
- m <- try(drm(response ~ dose, data = tmp, fct = BC.4(fixed = c(NA, 1, NA, NA))),
- silent = TRUE)
+ m <- try(drm(response ~ dose, data = tmp, fct = BC.4(fixed = c(NA, 1, NA, NA))))
if (chooseone==FALSE || fit==FALSE) {
if (!inherits(m, "try-error")) {
fit <- TRUE
@@ -91,26 +100,25 @@ drcfit <- function(data, chooseone=TRUE,
a[[ri]] <- coef(m)[[2]]
b[[ri]] <- coef(m)[[1]]
c[[ri]] <- coef(m)[[3]]
- ED50 <- try(ED(m, 50, interval = "delta",
+ ED50 <- try(ED(m, 50, interval = "delta",
lower = lowestdose / 10,
upper = highestdose * 10,
- display = FALSE),
- silent = TRUE)
+ display = FALSE))
if (!inherits(ED50, "try-error")) {
- logED50[[ri]] <- log10(ED50["1:50", "Estimate"])
- logED50low[[ri]] <- log10(ED50["1:50", "Lower"])
- logED50high[[ri]] <- log10(ED50["1:50", "Upper"])
+ logED50[[ri]] <- log10(ED50[ED50_row_index, "Estimate"])
+ logED50low[[ri]] <- log10(ED50[ED50_row_index, "Lower"])
+ logED50high[[ri]] <- log10(ED50[ED50_row_index, "Upper"])
if (logED50[[ri]] > rlhd[[ri]]) {
mtype[[ri]] <- "no fit"
}
- }
+ }
}
}
}
if (probit)
{
- m <- try(drm(response ~ dose, data = tmp,
- fct = LN.2()), silent = TRUE)
+ m <- try(drm(response ~ dose, data = tmp,
+ fct = LN.2()))
if (chooseone==FALSE || fit==FALSE) {
if (!inherits(m, "try-error")) {
fit <- TRUE
@@ -136,16 +144,15 @@ drcfit <- function(data, chooseone=TRUE,
} else {
mtype[[ri]] <- "probit"
ED50 <- ED(m, 50, interval = "delta", display = FALSE)
- logED50low[[ri]] <- log10(ED50["1:50", "Lower"])
- logED50high[[ri]] <- log10(ED50["1:50", "Upper"])
+ logED50low[[ri]] <- log10(ED50[ED50_row_index, "Lower"])
+ logED50high[[ri]] <- log10(ED50[ED50_row_index, "Upper"])
}
}
}
}
if (logit)
{
- m <- try(drm(response ~ dose, data = tmp, fct = LL.2()),
- silent = TRUE)
+ m <- try(drm(response ~ dose, data = tmp, fct = LL.2()))
if (chooseone==FALSE || fit==FALSE) {
if (!inherits(m, "try-error")) {
fit <- TRUE
@@ -173,8 +180,8 @@ drcfit <- function(data, chooseone=TRUE,
} else {
mtype[[ri]] <- "logit"
ED50 <- ED(m, 50, interval = "delta", display = FALSE)
- logED50low[[ri]] <- log10(ED50["1:50", "Lower"])
- logED50high[[ri]] <- log10(ED50["1:50", "Upper"])
+ logED50low[[ri]] <- log10(ED50[ED50_row_index, "Lower"])
+ logED50high[[ri]] <- log10(ED50[ED50_row_index, "Upper"])
}
}
}
@@ -182,8 +189,7 @@ drcfit <- function(data, chooseone=TRUE,
}
if (weibull)
{
- m <- try(drm(response ~ dose, data = tmp, fct = W1.2()),
- silent = TRUE)
+ m <- try(drm(response ~ dose, data = tmp, fct = W1.2()))
if (chooseone==FALSE || fit==FALSE) {
if (!inherits(m, "try-error")) {
fit <- TRUE
@@ -199,7 +205,7 @@ drcfit <- function(data, chooseone=TRUE,
rlhd[[ri]] <- log10(highestdose)
c[[ri]] <- NA
ED50 <- ED(m, 50, interval = "delta", display = FALSE)
- logED50[[ri]] <- log10(ED50["1:50", "Estimate"])
+ logED50[[ri]] <- log10(ED50[ED50_row_index, "Estimate"])
if (logED50[[ri]] > rlhd[[ri]]) {
mtype[[ri]] <- "no fit"
logED50[[ri]] <- NA
@@ -209,8 +215,8 @@ drcfit <- function(data, chooseone=TRUE,
b[[ri]] <- NA
} else {
mtype[[ri]] <- "weibull"
- logED50low[[ri]] <- log10(ED50["1:50", "Lower"])
- logED50high[[ri]] <- log10(ED50["1:50", "Upper"])
+ logED50low[[ri]] <- log10(ED50[ED50_row_index, "Lower"])
+ logED50high[[ri]] <- log10(ED50[ED50_row_index, "Upper"])
a[[ri]] <- logED50[[ri]]
b[[ri]] <- coef(m)[[1]]
}
@@ -256,7 +262,7 @@ drcfit <- function(data, chooseone=TRUE,
}
}
- results <- data.frame(rsubstance, rndl, rn, rlld, rlhd, mtype,
+ results <- data.frame(rsubstance, rndl, rn, rlld, rlhd, mtype,
logED50, logED50low, logED50high, runit, sigma, a, b)
lower_level_percent = paste(100 * (1 - level)/2, "%", sep = "")
upper_level_percent = paste(100 * (1 + level)/2, "%", sep = "")
@@ -279,14 +285,14 @@ drcfit <- function(data, chooseone=TRUE,
mtype <- as.character(results[row.i, "mtype"])
if (mtype %in% c("probit", "logit", "weibull", "linlogit")) {
for (EDi in EDx) {
- EDx.drc = try(ED(m, EDi, interval = "delta", display = FALSE, level = level),
- silent = TRUE)
+ EDx.drc = try(ED(m, EDi, interval = "delta", display = FALSE, level = level))
if (!inherits(EDx.drc, "try-error")) {
- results[row.i, paste0("EDx", EDi)] <- EDx.drc[paste0("1:", EDi), "Estimate"]
- results[row.i, paste0("EDx", EDi, " ", lower_level_percent)] <- EDx.drc[paste0("1:", EDi),
- "Lower"]
- results[row.i, paste0("EDx", EDi, " ", upper_level_percent)] <- EDx.drc[paste0("1:", EDi),
- "Upper"]
+ results[row.i, paste0("EDx", EDi)] <-
+ EDx.drc[paste(EDx_row_index_prefix, EDi, sep = ":"), "Estimate"]
+ results[row.i, paste0("EDx", EDi, " ", lower_level_percent)] <-
+ EDx.drc[paste(EDx_row_index_prefix, EDi, sep = ":"), "Lower"]
+ results[row.i, paste0("EDx", EDi, " ", upper_level_percent)] <-
+ EDx.drc[paste(EDx_row_index_prefix, EDi, sep = ":"), "Upper"]
}
}
}
diff --git a/R/drdata.R b/R/drdata.R
index dc741a6..0bf9597 100644
--- a/R/drdata.R
+++ b/R/drdata.R
@@ -21,7 +21,7 @@ drdata <- function(substances, experimentator = "%", db = "cytotox",
type <- organism
}
}
-
+
query <- paste("SELECT conc,",responsetype,",unit,experimentator,substance,",testtype,
",ok FROM ", db, " WHERE substance IN ('",
slist,"') AND experimentator LIKE '",
diff --git a/R/drfit.R b/R/drfit.R
index 65ab8e4..7fa53bf 100644
--- a/R/drfit.R
+++ b/R/drfit.R
@@ -18,7 +18,7 @@ drfit <- function(data, startlogED50 = NA, chooseone=TRUE,
# model result was appended
rsubstance <- array() # the substance names in the results
rndl <- vector() # number of dose levels
- rn <- vector() # mean number of replicates
+ rn <- vector() # mean number of replicates
# in each dose level
runit <- vector() # vector of units for each result row
rlhd <- rlld <- vector() # highest and lowest doses tested
@@ -75,13 +75,12 @@ drfit <- function(data, startlogED50 = NA, chooseone=TRUE,
active <- TRUE
} else {
active <- FALSE
- if (linlogit &&
+ if (linlogit &&
length(subset(linlogitWrong,linlogitWrong == i))==0 &&
length(subset(allWrong,allWrong == i))==0) {
m <- try(nls(response ~ linlogitf(dose,1,f,logED50,b),
- data=tmp, algorithm="port",
- start=list(f=f0,logED50=startlogED50[[i]],b=b0)),
- silent = TRUE)
+ data=tmp, algorithm="port",
+ start=list(f=f0,logED50=startlogED50[[i]],b=b0)))
if (!inherits(m, "try-error")) {
fit <- TRUE
ri <- ri + 1
@@ -105,8 +104,7 @@ drfit <- function(data, startlogED50 = NA, chooseone=TRUE,
c[[ri]] <- NA
} else {
mtype[[ri]] <- "linlogit"
- logED50conf <- try(confint(m,"logED50",level=level),
- silent = TRUE)
+ logED50conf <- try(confint(m,"logED50",level=level))
if (!inherits(logED50conf, "try-error")) {
logED50low[[ri]] <- logED50conf[[1]]
logED50high[[ri]] <- logED50conf[[2]]
@@ -124,9 +122,8 @@ drfit <- function(data, startlogED50 = NA, chooseone=TRUE,
if (probit &&
length(subset(allWrong,allWrong == i))==0) {
m <- try(nls(response ~ pnorm(-log10(dose),-logED50,scale),
- data=tmp, algorithm="port",
- start=list(logED50=startlogED50[[i]],scale=ps0)),
- silent = TRUE)
+ data=tmp, algorithm="port",
+ start=list(logED50=startlogED50[[i]],scale=ps0)))
if (chooseone==FALSE || fit==FALSE) {
if (!inherits(m, "try-error")) {
fit <- TRUE
@@ -151,8 +148,7 @@ drfit <- function(data, startlogED50 = NA, chooseone=TRUE,
b[[ri]] <- NA
} else {
mtype[[ri]] <- "probit"
- logED50conf <- try(confint(m,"logED50",level=level),
- silent = TRUE)
+ logED50conf <- try(confint(m,"logED50",level=level))
if (!inherits(logED50conf, "try-error")) {
logED50low[[ri]] <- logED50conf[[1]]
logED50high[[ri]] <- logED50conf[[2]]
@@ -170,9 +166,8 @@ drfit <- function(data, startlogED50 = NA, chooseone=TRUE,
if (logit &&
length(subset(allWrong,allWrong == i))==0) {
m <- try(nls(response ~ plogis(-log10(dose),-logED50,scale),
- data=tmp, algorithm="port",
- start=list(logED50=startlogED50[[i]],scale=ls0)),
- silent = TRUE)
+ data=tmp, algorithm="port",
+ start=list(logED50=startlogED50[[i]],scale=ls0)))
if (chooseone==FALSE || fit==FALSE) {
if (!inherits(m, "try-error")) {
fit <- TRUE
@@ -198,8 +193,7 @@ drfit <- function(data, startlogED50 = NA, chooseone=TRUE,
b[[ri]] <- NA
} else {
mtype[[ri]] <- "logit"
- logED50conf <- try(confint(m,"logED50",level=level),
- silent = TRUE)
+ logED50conf <- try(confint(m,"logED50",level=level))
if (!inherits(logED50conf, "try-error")) {
logED50low[[ri]] <- logED50conf[[1]]
logED50high[[ri]] <- logED50conf[[2]]
@@ -215,9 +209,8 @@ drfit <- function(data, startlogED50 = NA, chooseone=TRUE,
if (weibull &&
length(subset(allWrong,allWrong == i))==0) {
m <- try(nls(response ~ pweibull(-log10(dose)+location,shape),
- data=tmp, algorithm="port",
- start=list(location=startlogED50[[i]],shape=ws0)),
- silent = TRUE)
+ data=tmp, algorithm="port",
+ start=list(location=startlogED50[[i]],shape=ws0)))
if (chooseone==FALSE || fit==FALSE) {
if (!inherits(m, "try-error")) {
ri <- ri + 1
@@ -225,7 +218,7 @@ drfit <- function(data, startlogED50 = NA, chooseone=TRUE,
a[[ri]] <- coef(m)[["location"]]
b[[ri]] <- coef(m)[["shape"]]
sqrdev <- function(logdose) {
- (0.5 - pweibull( - logdose + a[[ri]], b[[ri]]))^2
+ (0.5 - pweibull( - logdose + a[[ri]], b[[ri]]))^2
}
logED50[[ri]] <- nlm(sqrdev,startlogED50[[i]])$estimate
if (sqrdev(logED50[[ri]]) > 0.1) {
@@ -298,7 +291,7 @@ drfit <- function(data, startlogED50 = NA, chooseone=TRUE,
c[[ri]] <- NA
}
}
- results <- data.frame(rsubstance, rndl, rn, rlld, rlhd, mtype,
+ results <- data.frame(rsubstance, rndl, rn, rlld, rlhd, mtype,
logED50, logED50low, logED50high, runit, sigma, a, b)
lower_level_percent = paste(100 * (1 - level)/2, "%", sep = "")
upper_level_percent = paste(100 * (1 + level)/2, "%", sep = "")
@@ -318,18 +311,18 @@ drfit <- function(data, startlogED50 = NA, chooseone=TRUE,
if (mtype[[row.i]] %in% c("probit", "logit", "weibull", "linlogit")) {
for (ED in EDx) {
of <- function(x) {
- abs(predict(models[[row.i]], data.frame(dose = 10^x)) -
+ abs(predict(models[[row.i]], data.frame(dose = 10^x)) -
(1 - (ED/100)))
- }
- # Search over interval starting an order of magnitude below
+ }
+ # Search over interval starting an order of magnitude below
# the lowest dose up to one order of magnitude above the
# highest dose
- o = optimize(of,
+ o = optimize(of,
results[row.i, c("lld", "lhd")] + c(-1, 1))
# Only keep results within the tolerance
if ((o$objective) < EDx.tolerance) {
logdose.ED = o$minimum
- results[row.i, paste0("EDx", ED)] <- 10^logdose.ED
+ results[row.i, paste0("EDx", ED)] <- 10^logdose.ED
}
}
}
diff --git a/R/drplot.R b/R/drplot.R
index 1e6b2cd..11c4d42 100644
--- a/R/drplot.R
+++ b/R/drplot.R
@@ -1,12 +1,12 @@
if(getRversion() >= '2.15.1') utils::globalVariables(c("dose", "Substance", "mtype"))
-drplot <- function(drresults, data,
+drplot <- function(drresults, data,
dtype = "std", alpha = 0.95, ctype = "none",
path = "./", fileprefix = "drplot", overlay = FALSE,
xlim = c("auto","auto"), ylim = c("auto","auto"),
xlab = paste("Decadic Logarithm of the dose in ", unit),
ylab = "Normalized response",
axes = TRUE, frame.plot = TRUE,
- postscript = FALSE, pdf = FALSE, png = FALSE,
+ postscript = FALSE, pdf = FALSE, png = FALSE,
bw = TRUE,
pointsize = 12,
colors = 1:8, ltys = 1:8, pchs = "auto",
@@ -24,7 +24,7 @@ drplot <- function(drresults, data,
if(is.data.frame(data)) {
# Get rid of pseudo substance names of controls
nonzerodata <- subset(data,dose!=0)
- nonzerodata$substance <- factor(nonzerodata$substance)
+ nonzerodata$substance <- factor(nonzerodata$substance)
zerodata <- subset(data,dose==0)
nc <- length(zerodata$dose) # Number of control points
if (nc > 0) {
@@ -44,12 +44,12 @@ drplot <- function(drresults, data,
hr <- max(nonzerodata$response)
if (ctype == "std") hr <- max(hr,1 + sdc)
if (ctype == "conf") hr <- max(hr,1 + controlconf)
- dsubstances <- levels(nonzerodata$substance)
+ dsubstances <- levels(nonzerodata$substance)
} else {
lld <- min(drresults[["logED50"]],na.rm=TRUE) - 2
lhd <- max(drresults[["logED50"]],na.rm=TRUE) + 2
if (length(subset(drresults,mtype=="linlogit")$Substance) != 0) {
- hr <- 1.8
+ hr <- 1.8
} else {
hr <- 1.0
}
@@ -70,20 +70,20 @@ drplot <- function(drresults, data,
postscript(file=filename,
paper="special",width=7,height=7,horizontal=FALSE, pointsize=pointsize)
message("Created File: ",filename,"\n")
- }
+ }
if (pdf) {
filename = paste(path,fileprefix,".pdf",sep="")
pdf(file=filename,
paper="special",width=7,height=7,horizontal=FALSE, pointsize=pointsize)
message("Created File: ",filename,"\n")
- }
+ }
if (png) {
filename = paste(path,fileprefix,".png",sep="")
png(filename=filename,
width=500, height=500, pointsize=pointsize)
message("Created File: ",filename,"\n")
}
-
+
plot(0,type="n",
xlim = xlim,
ylim = ylim,
@@ -97,7 +97,7 @@ drplot <- function(drresults, data,
if (!postscript && !png && !pdf && length(dsubstances) > 1) {
op <- par(ask=TRUE)
on.exit(par(op))
- }
+ }
}
# nl is the overall number of fits to draw by different line types
nl <- 0
@@ -123,20 +123,20 @@ drplot <- function(drresults, data,
postscript(file=filename,
paper="special",width=7,height=7,horizontal=FALSE,pointsize=pointsize)
message("Created File: ",filename,"\n")
- }
+ }
if (pdf) {
filename = paste(path,fileprefix,sub(" ","_",i),".pdf",sep="")
pdf(file=filename,
paper="special",width=7,height=7,horizontal=FALSE,pointsize=pointsize)
message("Created File: ",filename,"\n")
- }
+ }
if (png) {
filename = paste(path,fileprefix,sub(" ","_",i),".png",sep="")
png(filename=filename,
width=500, height=500, pointsize=pointsize)
message("Created File: ",filename,"\n")
}
-
+
plot(0,type="n",
xlim = xlim,
ylim = ylim,
@@ -147,7 +147,7 @@ drplot <- function(drresults, data,
}
if (!overlay) legend(lpos, i, lty = 1, col = color, pch = pch, inset=0.05)
tmp$dosefactor <- factor(tmp$dose) # necessary because the old
- # factor has all levels, not
+ # factor has all levels, not
# only the ones tested with
# this substance
@@ -157,7 +157,7 @@ drplot <- function(drresults, data,
abline(h = 1 + sdc, lty = 2)
}
if (ctype == "conf") {
- abline(h = 1 - controlconf, lty = 2)
+ abline(h = 1 - controlconf, lty = 2)
abline(h = 1 + controlconf, lty = 2)
}
@@ -179,7 +179,7 @@ drplot <- function(drresults, data,
}
if (dtype == "conf")
{
- confidencedeltas <- qt((1 + alpha)/2, lengths - 1) * sqrt(vars)
+ confidencedeltas <- qt((1 + alpha)/2, lengths - 1) * sqrt(vars)
tops <- means + confidencedeltas
bottoms <- means - confidencedeltas
}
@@ -207,13 +207,13 @@ drplot <- function(drresults, data,
if (drresults[[fit.row, "mtype"]] %in% c("probit",
"logit",
"weibull",
- "linlogit"))
+ "linlogit"))
{
m <- attr(drresults, "models")[[as.numeric(fit.row)]]
of <- function(x) {
predict(m, data.frame(dose = 10^x))
- }
- plot(of, lld - 0.5, lhd + 2,
+ }
+ plot(of, lld - 0.5, lhd + 2,
add = TRUE, col = color, lty = lty)
}
}
diff --git a/data/XY.rda b/data/XY.rda
index 4b0d426..a4f6242 100644
--- a/data/XY.rda
+++ b/data/XY.rda
Binary files differ
diff --git a/inst/doc/drfit-Rnews.pdf b/inst/doc/drfit-Rnews.pdf
index 65ad25f..0eb0de4 100644
--- a/inst/doc/drfit-Rnews.pdf
+++ b/inst/doc/drfit-Rnews.pdf
Binary files differ

Contact - Imprint