"NaNs produced" 使用 drc 包计算绝对 EC50 值时出现警告
"NaNs produced" warning when calculating absolute EC50 values with drc package
我正在尝试弄清楚如何使用 drc 包中的 LL.3 和 LL.4（3 和 4 参数）剂量反应模型计算绝对 EC50 值，但我不断收到 "Warning message:In log(exp(-tempVal/parmVec[5]) - 1) : NaNs produced" 且 EC50 值为 "NA".
这是我正在尝试的代码示例 运行
###use rygrass dataset in drc
gr.LL.3 <- drm(ryegrass$rootl ~ ryegrass$conc, fct = LL.3()) # 3 parameter log-logistic model
gr.LL.4 <- drm(ryegrass$rootl ~ ryegrass$conc, fct = LL.4()) # 4 parameter log-logistic model
plot(gr.LL.3) #graph looks fine
plot(gr.LL.4) #graph looks fine
ED (gr.LL.3, respLev = c(50), type = "relative") # this works fine
ED (gr.LL.4, respLev = c(50), type = "relative") # this works fine
ED (gr.LL.3, respLev = c(50), type = "absolute") # this gives me "NA" for EC50 along with warning message
ED (gr.LL.4, respLev = c(50), type = "absolute") # this gives me "NA" for EC50 along with warning message
这不是因为浓度值为 0
### It's not due to 0 values for concentrations
# ryegrass dataset with 0 value concentrations and corresponding rootl removed
rootlength <- c(8.3555556, 6.9142857, 7.75, 6.8714286, 6.45, 5.9222222, 1.925, 2.8857143, 4.2333333, 1.1875, 0.8571429, 1.0571429, 0.6875, 0.525, 0.825, 0.25, 0.22, 0.44)
conc.wo.0 <- c(0.94, 0.94, 0.94, 1.88, 1.88, 1.88, 3.75, 3.75, 3.75, 7.5, 7.5, 7.5, 15, 15, 15, 30, 30, 30)
gro.LL.3 <- drm(rootlength ~ conc.wo.0, fct = LL.3())
plot(gro.LL.3) #graph looks fine
ED (gro.LL.3, respLev = c(50), type = "relative") # this works fine
ED (gro.LL.3, respLev = c(50), type = "absolute") # once again, this gives me "NA" for EC50 along with warning message
这也不是因为响应是绝对的还是相对的
### It's also not due to the response being in absolute vs relative terms
# ryegrass dataset with response relative to average response with 0 concentration (sorry, I did the absolute to relative conversion in excel, I'm still learning r)
rel.rootl <- c(0.98, 1.03, 1.07, 0.94, 0.95, 1.03, 1.08, 0.89, 1.00, 0.89, 0.83, 0.76, 0.25, 0.37, 0.55, 0.15, 0.11, 0.14, 0.09, 0.07, 0.11, 0.03, 0.03, 0.06)
concentration <- c(0, 0, 0, 0, 0, 0, 0.94, 0.94, 0.94, 1.88, 1.88, 1.88, 3.75, 3.75, 3.75, 7.5, 7.5, 7.5, 15, 15, 15, 30, 30, 30)
rel.gro.LL.3 <- drm(rel.rootl ~ concentration, fct = LL.3())
plot(rel.gro.LL.3) #graph looks fine
ED (rel.gro.LL.3, respLev = c(50), type = "relative") # this works fine
ED (rel.gro.LL.3, respLev = c(50), type = "absolute") # once again, this gives me "NA" for EC50 along with warning message
我是新手，所以非常感谢您的帮助。
rel.rootl <- c(0.98, 1.03, 1.07, 0.94, 0.95, 1.03, 1.08, 0.89, 1.00, 0.89, 0.83, 0.76, 0.25, 0.37, 0.55, 0.15, 0.11, 0.14, 0.09, 0.07, 0.11, 0.03, 0.03, 0.06)
concentration <- c(0, 0, 0, 0, 0, 0, 0.94, 0.94, 0.94, 1.88, 1.88, 1.88, 3.75, 3.75, 3.75, 7.5, 7.5, 7.5, 15, 15, 15, 30, 30, 30)
rel.gro.LL.3 <- drm(rel.rootl ~ concentration, fct = LL.3())
plot(rel.gro.LL.3) #graph looks fine
ED (rel.gro.LL.3, respLev = c(50), type = "relative") # this works fine
ED (rel.gro.LL.3, respLev = c(50), type = "absolute") # once again, this gives me "NA" for EC50 along with warning message
问题是因为当您尝试估计绝对 EC50 时，ED 函数求解曲线上您想要的点（即 respLev 参数），因此如果您的相对响应水平在y 轴它会 运行 出错，因为你的 y 轴是比例。
要解决此问题，请将您的标准化响应乘以 100，将其转换为相对响应百分比
rel.gro.LL.3.percent <- drm(rel.rootl*100 ~ concentration, fct = LL.3())
ED (rel.gro.LL.3.percent, respLev = c(50), type = "relative") # same result as above
Estimated effective doses
Estimate Std. Error
e:1:50 3.26520 0.19915
ED (rel.gro.LL.3.percent, respLev = c(50), type = "absolute") # very similar to relative EC50
Estimated effective doses
Estimate Std. Error
e:1:50 3.30154 0.20104
或者，您可以将原始模型中的 respLev 更改为 0.5。
ED (rel.gro.LL.3, respLev = c(50), type = "relative") # this still works fine
Estimated effective doses
Estimate Std. Error
e:1:50 3.26520 0.19915
ED (rel.gro.LL.3, respLev = c(0.5), type = "absolute") # Now this works and is the same as we got before with response multiplied by 100
Estimated effective doses
Estimate Std. Error
e:1:0.5 3.30154 0.20104
我正在尝试弄清楚如何使用 drc 包中的 LL.3 和 LL.4（3 和 4 参数）剂量反应模型计算绝对 EC50 值，但我不断收到 "Warning message:In log(exp(-tempVal/parmVec[5]) - 1) : NaNs produced" 且 EC50 值为 "NA".
这是我正在尝试的代码示例 运行
###use rygrass dataset in drc
gr.LL.3 <- drm(ryegrass$rootl ~ ryegrass$conc, fct = LL.3()) # 3 parameter log-logistic model
gr.LL.4 <- drm(ryegrass$rootl ~ ryegrass$conc, fct = LL.4()) # 4 parameter log-logistic model
plot(gr.LL.3) #graph looks fine
plot(gr.LL.4) #graph looks fine
ED (gr.LL.3, respLev = c(50), type = "relative") # this works fine
ED (gr.LL.4, respLev = c(50), type = "relative") # this works fine
ED (gr.LL.3, respLev = c(50), type = "absolute") # this gives me "NA" for EC50 along with warning message
ED (gr.LL.4, respLev = c(50), type = "absolute") # this gives me "NA" for EC50 along with warning message
这不是因为浓度值为 0
### It's not due to 0 values for concentrations
# ryegrass dataset with 0 value concentrations and corresponding rootl removed
rootlength <- c(8.3555556, 6.9142857, 7.75, 6.8714286, 6.45, 5.9222222, 1.925, 2.8857143, 4.2333333, 1.1875, 0.8571429, 1.0571429, 0.6875, 0.525, 0.825, 0.25, 0.22, 0.44)
conc.wo.0 <- c(0.94, 0.94, 0.94, 1.88, 1.88, 1.88, 3.75, 3.75, 3.75, 7.5, 7.5, 7.5, 15, 15, 15, 30, 30, 30)
gro.LL.3 <- drm(rootlength ~ conc.wo.0, fct = LL.3())
plot(gro.LL.3) #graph looks fine
ED (gro.LL.3, respLev = c(50), type = "relative") # this works fine
ED (gro.LL.3, respLev = c(50), type = "absolute") # once again, this gives me "NA" for EC50 along with warning message
这也不是因为响应是绝对的还是相对的
### It's also not due to the response being in absolute vs relative terms
# ryegrass dataset with response relative to average response with 0 concentration (sorry, I did the absolute to relative conversion in excel, I'm still learning r)
rel.rootl <- c(0.98, 1.03, 1.07, 0.94, 0.95, 1.03, 1.08, 0.89, 1.00, 0.89, 0.83, 0.76, 0.25, 0.37, 0.55, 0.15, 0.11, 0.14, 0.09, 0.07, 0.11, 0.03, 0.03, 0.06)
concentration <- c(0, 0, 0, 0, 0, 0, 0.94, 0.94, 0.94, 1.88, 1.88, 1.88, 3.75, 3.75, 3.75, 7.5, 7.5, 7.5, 15, 15, 15, 30, 30, 30)
rel.gro.LL.3 <- drm(rel.rootl ~ concentration, fct = LL.3())
plot(rel.gro.LL.3) #graph looks fine
ED (rel.gro.LL.3, respLev = c(50), type = "relative") # this works fine
ED (rel.gro.LL.3, respLev = c(50), type = "absolute") # once again, this gives me "NA" for EC50 along with warning message
我是新手，所以非常感谢您的帮助。
rel.rootl <- c(0.98, 1.03, 1.07, 0.94, 0.95, 1.03, 1.08, 0.89, 1.00, 0.89, 0.83, 0.76, 0.25, 0.37, 0.55, 0.15, 0.11, 0.14, 0.09, 0.07, 0.11, 0.03, 0.03, 0.06)
concentration <- c(0, 0, 0, 0, 0, 0, 0.94, 0.94, 0.94, 1.88, 1.88, 1.88, 3.75, 3.75, 3.75, 7.5, 7.5, 7.5, 15, 15, 15, 30, 30, 30)
rel.gro.LL.3 <- drm(rel.rootl ~ concentration, fct = LL.3())
plot(rel.gro.LL.3) #graph looks fine
ED (rel.gro.LL.3, respLev = c(50), type = "relative") # this works fine
ED (rel.gro.LL.3, respLev = c(50), type = "absolute") # once again, this gives me "NA" for EC50 along with warning message
问题是因为当您尝试估计绝对 EC50 时，ED 函数求解曲线上您想要的点（即 respLev 参数），因此如果您的相对响应水平在y 轴它会 运行 出错，因为你的 y 轴是比例。
要解决此问题，请将您的标准化响应乘以 100，将其转换为相对响应百分比
rel.gro.LL.3.percent <- drm(rel.rootl*100 ~ concentration, fct = LL.3())
ED (rel.gro.LL.3.percent, respLev = c(50), type = "relative") # same result as above
Estimated effective doses
Estimate Std. Error
e:1:50 3.26520 0.19915
ED (rel.gro.LL.3.percent, respLev = c(50), type = "absolute") # very similar to relative EC50
Estimated effective doses
Estimate Std. Error
e:1:50 3.30154 0.20104
或者，您可以将原始模型中的 respLev 更改为 0.5。
ED (rel.gro.LL.3, respLev = c(50), type = "relative") # this still works fine
Estimated effective doses
Estimate Std. Error
e:1:50 3.26520 0.19915
ED (rel.gro.LL.3, respLev = c(0.5), type = "absolute") # Now this works and is the same as we got before with response multiplied by 100
Estimated effective doses
Estimate Std. Error
e:1:0.5 3.30154 0.20104