GitHub - wvictor14/planet: For inferring ancestry, gestational age, cell type proportions, and preeclampsia status from placental DNA methylation array data (original) (raw)

planet

planet is an R package for inferring ethnicity (1),gestational age (2), cell composition (3), and preeclampsia (4), from placental DNA methylation data.

See full documentation atvictoryuan.com/planet

Installation

Latest Bioconductor release

if(!requireNamespace("BiocManager", quietly = TRUE)) install.packages("BiocManager") BiocManager::install("planet")

Or the development version of planet:

devtools::install_github('wvictor14/planet')

Usage

See vignettes for more detailed usage.

Example Data

All functions in this package take as input DNAm data the 450k and EPIC DNAm microarray. For best performance I suggest providing unfiltered data normalized with noob and BMIQ. A processed example dataset,plBetas, is provided to show the format that this data should be in. The output of all planet functions is a data.frame.

A quick example of each major function is illustrated with this example data:

library(minfi) library(planet)

#load example data data(plBetas) data(plPhenoData) # sample information

Predict Ethnicity

predictEthnicity(plBetas) |> head() #> 1860 of 1860 predictors present. #> # A tibble: 6 × 7 #> Sample_ID Predicted_ethnicity_n…¹ Predicted_ethnicity Prob_African Prob_Asian #> #> 1 GSM1944936 Caucasian Caucasian 0.00331 0.0164
#> 2 GSM1944939 Caucasian Caucasian 0.000772 0.000514 #> 3 GSM1944942 Caucasian Caucasian 0.000806 0.000699 #> 4 GSM1944944 Caucasian Caucasian 0.000883 0.000792 #> 5 GSM1944946 Caucasian Caucasian 0.000885 0.00130 #> 6 GSM1944948 Caucasian Caucasian 0.000852 0.000973 #> # ℹ abbreviated name: ¹Predicted_ethnicity_nothresh #> # ℹ 2 more variables: Prob_Caucasian , Highest_Prob

Predict Gestational Age

There are 3 gestational age clocks for placental DNA methylation data from Lee Y. et al. 2019 (2). To use a specific one, we can use thetype argument in predictAge:

predictAge(plBetas, type = 'RPC') |> head() #> 558 of 558 predictors present. #> [1] 38.46528 33.09680 34.32520 35.50937 37.63910 36.77051

Predict Cell Composition

Reference data to infer cell composition on placental villi DNAm samples (3) can be used with cell deconvolution from minfi or EpiDISH. These are provided in this package as plCellCpGsThird and plCellCpGsFirst for third trimester (term) and first trimester samples, respectively.

data('plCellCpGsThird')

minfi:::projectCellType(

subset your data to cell cpgs

plBetas[rownames(plCellCpGsThird),],

input the reference cpg matrix

plCellCpGsThird,

lessThanOne = FALSE) |>

head() #> Trophoblasts Stromal Hofbauer Endothelial nRBC #> GSM1944936 0.1091279 0.04891919 0.000000e+00 0.08983998 0.05294062 #> GSM1944939 0.2299918 0.00000000 6.680983e-20 0.07888007 0.03374149 #> GSM1944942 0.1934287 0.03483540 0.000000e+00 0.09260353 0.02929310 #> GSM1944944 0.2239896 0.06249135 1.608645e-03 0.11040693 0.04447951 #> GSM1944946 0.1894152 0.07935955 0.000000e+00 0.10587439 0.05407587 #> GSM1944948 0.2045124 0.07657717 0.000000e+00 0.09871149 0.02269798 #> Syncytiotrophoblast #> GSM1944936 0.6979477 #> GSM1944939 0.6377822 #> GSM1944942 0.6350506 #> GSM1944944 0.5467642 #> GSM1944946 0.6022329 #> GSM1944948 0.6085825

Predict Preeclampsia

download the model from experimenthub

library(ExperimentHub) #> Loading required package: AnnotationHub #> Loading required package: BiocFileCache #> Loading required package: dbplyr #> #> Attaching package: 'AnnotationHub' #> The following object is masked from 'package:Biobase': #> #> cache eh <- ExperimentHub()

query(eh, "eoPredData") # see data

download BMIQ normalized 450k data for prediction

x_test <- eh[['EH8403']] #> see ?eoPredData and browseVignettes('eoPredData') for documentation #> loading from cache preds <- x_test |> predictPreeclampsia() #> see ?eoPredData and browseVignettes('eoPredData') for documentation #> loading from cache #> 45 of 45 predictive CpGs present. #> BMIQ normalization is recommended for best results. If choosing other method, it is recommended to compare results to predictions on BMIQ normalized data. preds |> head() #> # A tibble: 6 × 4 #> Sample_ID EOPE Non-PE Preterm PE_Status
#>
#> 1 GSM2589533 0.670 0.330 EOPE
#> 2 GSM2589535 0.768 0.232 EOPE
#> 3 GSM2589536 0.807 0.193 EOPE
#> 4 GSM2589538 0.784 0.216 EOPE
#> 5 GSM2589540 0.386 0.614 Normotensive #> 6 GSM2589541 0.649 0.351 EOPE