-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathREADME.Rmd
155 lines (104 loc) · 5.59 KB
/
README.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
---
output: github_document
---
<!-- README.md is generated from README.Rmd. Please edit that file -->
```{r, include = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.path = "man/figures/README-",
out.width = "100%"
)
```
# pksensi <img src="man/figures/logo.png" height="200px" align="right" alt="pksensi logo" />
<!-- badges: start -->
[](https://www.tidyverse.org/lifecycle/#experimental)
[](https://github.com/nanhung/pksensi/actions)
[](https://codecov.io/gh/nanhung/pksensi?branch=master)
<!-- badges: end -->
**pksensi** implements the global sensitivity analysis workflow to investigate the parameter uncertainty and sensitivity in physiologically based kinetic (PK) models, especially the physiologically based pharmacokinetic/toxicokinetic model with multivariate outputs. The package also provides some functions to check the convergence and sensitivity of model parameters.
Through **pksensi**, you can:
- Run sensitivity analysis for PK models in R with script that were written in C or GNU MCSim.
- Decision support: The output results and visualization tools can be used to easily determine which parameters have "non-influential" effects on the model output and can be fixed in following model calibration.
## Installation
You can install the released version of **pksensi** from [CRAN](https://CRAN.R-project.org) with:
``` r
install.packages("pksensi")
```
And the development version from [GitHub](https://github.com/) with:
``` r
# install.packages("remotes")
remotes::install_github("nanhung/pksensi")
```
- This package includes a function that can help you install GNU MCSim more easily through the function `mcsim_install()`.
- All updated details can be found in [NEWS.md](https://github.com/nanhung/pksensi/blob/master/NEWS.md).
- **NOTE:** Windows users need to install [Rtools40](https://cran.r-project.org/bin/windows/Rtools/) to compile the model code.
## Workflow
<img src="man/figures/sensitivity-workflow.png" align="left" alt="Workflow of sensitivity analysis" />
**Note:** The parameter correlation (e.g., V~max~ and K~M~ in metabolism) might be an issue in the global sensitivity analysis. If you have experiment data, suggest using small datasets as a sample in Markov Chain Monte Carlo Simulation. Then, check correlation before conducting the sensitivity analysis. The issue will try to address in the future version.
## Example
This is a basic example of applying **pksensi** in one-compartment pbtk model:
```{r example}
library(pksensi)
```
### Step 1. Construct 1-cpt pbtk model
```{r}
pbtk1cpt <- function(t, state, parameters) {
with(as.list(c(state, parameters)), {
dAgutlument = - kgutabs * Agutlument
dAcompartment = kgutabs * Agutlument - ke * Acompartment
dAmetabolized = ke * Acompartment
Ccompartment = Acompartment / vdist * BW;
list(c(dAgutlument, dAcompartment, dAmetabolized),
"Ccompartment" = Ccompartment)
})
}
```
### Step 2. Define initial conditions, output time steps and variable
```{r}
initState <- c(Agutlument = 10, Acompartment = 0, Ametabolized = 0)
t <- seq(from = 0.01, to = 24.01, by = 1)
outputs <- c("Ccompartment")
```
### Step 3. Generate parameter matrix
#### 3.1. (Optional) Extract parameter value from httk package
```{r, message=FALSE, warning=FALSE}
library(httk)
pars1comp <- (parameterize_1comp(chem.name = "acetaminophen"))
```
#### 3.2. Set parameter distributions
```{r}
q <- c("qunif", "qunif", "qunif", "qnorm")
q.arg <- list(list(min = pars1comp$Vdist / 2, max = pars1comp$Vdist * 2),
list(min = pars1comp$kelim / 2, max = pars1comp$kelim * 2),
list(min = pars1comp$kgutabs / 2, max = pars1comp$kgutabs * 2),
list(mean = pars1comp$BW, sd = 5))
```
#### 3.3. Create parameter matrix
```{r}
set.seed(1234)
params <- c("vdist", "ke", "kgutabs", "BW")
x <- rfast99(params, n = 200, q = q, q.arg = q.arg, replicate = 1)
```
### Step 4. Conduct simulation (will take few minutes with more replications)
```{r}
out <- solve_fun(x, time = t, func = pbtk1cpt, initState = initState, outnames = outputs)
```
### Step 5. Uncertainty analysis
```{r fig.alt="pksim plot"}
pksim(out) # Use to compare with "real" data (if any)
```
### Step 6. Check and visualize the result of sensitivity analysis
```{r fig.alt="autoplot"}
plot(out) # Visualize result
check(out) # Print result to console
```
## Citation
```{r, comment = "", echo = FALSE}
citation(package = "pksensi")
```
## Reference
Hsieh NH, Reisfeld B, Bois FY, Chiu WA. [Applying a global sensitivity analysis workflow to improve the computational efficiencies in physiologically-based pharmacokinetic modeling](https://www.frontiersin.org/articles/10.3389/fphar.2018.00588/full). Frontiers in Pharmacology 2018 Jun; 9:588.
Hsieh NH, Reisfeld B, Chiu WA. [pksensi: An R package to apply global sensitivity analysis in physiologically based kinetic modeling](https://doi.org/10.1016/j.softx.2020.100609). SoftwareX 2020 Jul; 12:100609.
Hsieh NH, Bois FY, Tsakalozou E, Ni Z, Yoon M, Sun W, Klein M, Reisfeld B, Chiu WA. [A Bayesian population physiologically based pharmacokinetic absorption modeling approach to support generic drug development: application to bupropion hydrochloride oral dosage forms](https://doi.org/10.1007/s10928-021-09778-5). Journal of Pharmacokinetics and Pharmacodynamics 2021 Sep; 22:1-6.