Introduction to scPred
Jose Alquicira Hernandez
2018-07-24
Summary
scPred is a general method to predict cell types based on variance structure decomposition. It selects the most cell type-informative principal components from a dataset and trains a prediction model for each cell type. The principal training axes are projected onto the test dataset to obtain the PCs scores for the test dataset and the trained model(s) is/are used to classify single cells.
For more details see our pre-print on bioRxiv:
Application of scPred
First, we load the scPred package and tidyverse.
library("scPred")
library("tidyverse")We will work with single cell data of pluripotent, blood, skin and neural cells sequenced at low coverage. For more details about the study, see Low-coverage single-cell mRNA sequencing reveals cellular heterogeneity and activated signaling pathways in developing cerebral cortex.
The count matrix and metadata may be obtained here from Hemberg’s lab.
Read the gene expression data (SingleCellExperiment object), calculate CPM values and extract metadata.
download.file("https://scrnaseq-public-datasets.s3.amazonaws.com/scater-objects/pollen.rds", destfile = "~/Downloads/pollen.rds")require(SingleCellExperiment)
pollen <- readRDS("~/Downloads/pollen.rds")pollen_counts <- normcounts(pollen)
pollen_cpm <- apply(pollen_counts, 2, function(x) (x/sum(x))*1000000)
pollen_metadata <- as.data.frame(colData(pollen))Let’s explore the cell type information
table(pollen_metadata$cell_type2)##
## blood dermal neural pluripotent
## 113 99 65 24A total of 301 cells are included in the dataset.
For demostration purposes, we split a gene expression matrix into two groups (train and test datasets) based cell type information using the createDataPartition() function from the caret package (already loaded with scPred).
The train partition will be used to train a prediction models for each cell type and finally, the models will be tested using the test partition.
set.seed(1234)
i <- createDataPartition(pollen_metadata$cell_type2, p = 0.70, list = FALSE)
train_data <- t(pollen_cpm[, i])
test_data <- t(pollen_cpm[, -i])
train_info <- pollen_metadata[i, , drop = FALSE]
test_info <- pollen_metadata[-i, , drop = FALSE]Training step
Eigendecomposition
The first part of the scPred algorithm consists on decomposing the gene expresion matrix of the training dataset to obtained a low dimensional space that can describe most of the variance of the dataset. The eigenDecompose function calculates the first n principal components and log-transforms the input gene expression values to stabilize the variance. It returns an scPred object.
set.seed(1234)
scp <- eigenDecompose(train_data, n = 10)Then, we assign the metadata containing the cell type information. Row names in the metadata dataframe must match the row names from the eigendecompsed gene expression matrix.
scPred::metadata(scp) <- train_infoFeature selection
Next, we select the principal components that explain the class identity of each cell type using the getFeatureSpace function. This function applies a Wilcoxcon rank sum test to determine the informative principal components according to a categorical variable variable. In this case, we want to predict the cell types in the cell_type2 columns from the metadata. Run ?getFeatureSpace for more details.
scp <- getFeatureSpace(scp, pVar = "cell_type2")## DONE!The features slot contains the principal components that explain the class identity.
- pValue contains the associated p-value for each principal component obtained using the Wilcoxon Rank sum test
- pValueAdj is the adjusted p-value depending omn the correction criterion applied. By defauls a false discovery rate corrections is performed
- expVar contrains the explained variance by each principal component
- cumExpVar contains the cumulative variance explained
All prrincipal components for each cell type are ranked by p-value.
scp@features## $blood
## PC pValue pValueAdj expVar cumExpVar
## 1 PC2 6.186072e-26 6.186072e-25 20.291116 20.29112
## 2 PC5 3.594918e-10 1.797459e-09 8.573597 28.86471
## 3 PC3 2.822222e-08 9.407408e-08 11.433759 40.29847
## 4 PC4 2.866624e-07 7.166560e-07 10.080527 50.37900
##
## $dermal
## PC pValue pValueAdj expVar cumExpVar
## 1 PC2 1.378227e-29 1.378227e-28 20.29112 20.29112
## 2 PC1 1.300270e-09 6.501350e-09 26.21813 46.50925
## 3 PC4 7.311207e-08 2.437069e-07 10.08053 56.58977
## 4 PC6 2.872213e-05 7.180532e-05 7.26231 63.85208
## 5 PC3 2.034155e-04 4.068310e-04 11.43376 75.28584
##
## $neural
## PC pValue pValueAdj expVar cumExpVar
## 1 PC1 9.513036e-21 9.513036e-20 26.21813 26.21813
## 2 PC4 1.103454e-15 5.517270e-15 10.08053 36.29866
## 3 PC3 7.019078e-05 2.339693e-04 11.43376 47.73242
##
## $pluripotent
## PC pValue pValueAdj expVar cumExpVar
## 1 PC6 8.590373e-12 8.590373e-11 7.262310 7.26231
## 2 PC4 2.420729e-10 1.210364e-09 10.080527 17.34284
## 3 PC5 1.797335e-08 5.991115e-08 8.573597 25.91643
## 4 PC9 1.169709e-07 2.924273e-07 3.512692 29.42913
## 5 PC8 9.169523e-05 1.833905e-04 4.313980 33.74311
## 6 PC10 4.675783e-03 7.792971e-03 3.418549 37.16165
## 7 PC3 1.065838e-02 1.522625e-02 11.433759 48.59541
## 8 PC2 1.376060e-02 1.720075e-02 20.291116 68.88653We can plot the principal components grouped by the prediction variable using the plotEigen() function
plotEigen(scp, group = "cell_type2")
Model training
We can now train prediction models for blood, dermal, neural, and pluripotent cell types.
scp <- trainModel(scp, seed = 66)If we print the scPred object we can look at a summary of the slots contained in it.
- Expression data: shows the number of cells, genes, and principal components computed.
- Metadata information: Show the columns in the metadata slot. If columns are factor objects, they can be used as response veriables to train a prediction model
- Prediction: Shows the prediction variable as indicated using the
getFeatureSpace()function - Informative PCs per class: shows the number of discriminant principal components for each class (e.g. cell type)
- Training: Shows the description of the classification model used for training. For each class, performance metrics such as AUROC,accuracym or kappa are shown
The four models showed a specificity of 1 and a sensitivity of 0.99 to 1.
scp## 'scPred' object
## - Expression data
## Cells = 213
## Genes = 21413
## PCs = 10
## - Metadata information
## cell_type1, cell_type2
## - Prediction
## Variable = cell_type2
## - Informative PCs per class
## blood = 4
## dermal = 5
## neural = 3
## pluripotent = 8
## - Training
## Model: Support Vector Machines with Radial Basis Function Kernel
## Class -> blood
## AUROC = 1, Sensitivity = 0.988, Specificity = 1
## Class -> dermal
## AUROC = 1, Sensitivity = 1, Specificity = 1
## Class -> neural
## AUROC = 1, Sensitivity = 1, Specificity = 1
## Class -> pluripotent
## AUROC = 1, Sensitivity = 1, Specificity = 1We can plot the distribution of probabilities to see the performance of the predictions for each cell class using
The getTrainResults() function extracts the predictions results obtained from the resampling step for training the prediction model.
res <- getTrainResults(scp)We can plot the calculated probabilities for each cell type versus our cell labels:
mapply(function(x,y){dplyr::rename(x, probability = !! enquo(y))}, res, names(res), SIMPLIFY = FALSE) %>%
Reduce(rbind, .) -> train_probabilities
train_probabilities %>%
select(object, obs, probability, "other") %>%
ggplot() +
aes(probability, fill = obs) +
geom_histogram(bins = 30, color = "black") +
geom_vline(xintercept = 0.9, color = "red") +
facet_wrap(~object) +
theme_bw()
In the previous figure we can observe that a threshold of 0.9 classifies all dermal, neural and pluripotent cells correctly and almost all blood cells too. Each panel represents a prediction model and the colors the known true classes. All other cells are cells except the positive class (for example, for the blood prediction model all other cells are either dermal, neural, or pluripotent)
Prediction step
Once the models have been trained they can be applied to predict cell types in other dataset, for this demonstration we’ll use the test partition/ scPredict() projects the training principal axes onto the test dataset and predicts the cell identity using the trained models. By default, scPredict() uses a threshold of 0.9 to classify the cells into categories.
predictions <- scPredict(scp, newData = test_data, threshold = 0.9)scPredict() returns a dataframe with the probabilities of each cell to belong to any of the cell classes. The predClass columns is set using the provided threshold.
predictions## blood dermal neural pluripotent predClass
## Hi_2338_1 0.006655574 0.998660453 0.007440833 0.005008939 dermal
## Hi_2338_2 0.005973183 0.997339365 0.007017599 0.004994127 dermal
## Hi_2338_4 0.011011438 0.993270822 0.009921552 0.004863395 dermal
## Hi_2338_5 0.017761575 0.995914310 0.006822524 0.003790678 dermal
## Hi_2338_17 0.024224216 0.902759744 0.014851823 0.005115090 dermal
## Hi_2339_7 0.993408005 0.012042407 0.008148723 0.005297782 blood
## Hi_2339_8 0.997380247 0.007232218 0.005281070 0.005405074 blood
## Hi_2339_9 0.964734320 0.006105625 0.013411268 0.005270282 blood
## Hi_2339_11 0.993005444 0.007104105 0.004966572 0.005266879 blood
## Hi_2339_14 0.965739837 0.038056046 0.006520396 0.005291451 blood
## Hi_K562_2 0.995894429 0.005652630 0.002859333 0.003934111 blood
## Hi_K562_4 0.989691640 0.007281384 0.009687081 0.004711595 blood
## Hi_K562_10 0.988013653 0.006033125 0.002608131 0.004933672 blood
## Hi_BJ_1 0.010147845 0.995365243 0.005814412 0.003934786 dermal
## Hi_BJ_2 0.007979329 0.990617637 0.009042959 0.005142810 dermal
## Hi_BJ_6 0.009356208 0.995618949 0.008309160 0.005081485 dermal
## Hi_BJ_8 0.010222853 0.993650077 0.006426775 0.005213459 dermal
## Hi_BJ_10 0.007288164 0.996559180 0.005883454 0.004426737 dermal
## Hi_BJ_11 0.007089321 0.992122295 0.007084840 0.005234914 dermal
## Hi_BJ_12 0.007945532 0.995737015 0.005083199 0.004576097 dermal
## Hi_BJ_17 0.008009722 0.994678439 0.007570398 0.005267832 dermal
## Hi_BJ_19 0.007158134 0.989414384 0.008616647 0.005318494 dermal
## Hi_BJ_30 0.008399897 0.996345448 0.006304303 0.004977812 dermal
## Hi_BJ_34 0.009390083 0.996917264 0.007515548 0.003792042 dermal
## Hi_K562_19 0.994617229 0.006784354 0.009251010 0.004032288 blood
## Hi_K562_20 0.953711538 0.008983164 0.004315662 0.004139242 blood
## Hi_K562_22 0.994391053 0.005913928 0.004323663 0.004350294 blood
## Hi_K562_23 0.996412632 0.005749871 0.005622456 0.004688627 blood
## Hi_K562_25 0.994925606 0.005923280 0.004797062 0.004737113 blood
## Hi_K562_31 0.996464299 0.005393526 0.003702113 0.004653214 blood
## Hi_K562_34 0.995227159 0.006150064 0.005739733 0.005181328 blood
## Hi_K562_36 0.996773580 0.005926799 0.006433327 0.004129075 blood
## Hi_K562_38 0.995700707 0.005717990 0.003582286 0.004625931 blood
## Hi_K562_40 0.995912496 0.005283276 0.002926409 0.004131742 blood
## Hi_K562_41 0.996614506 0.005355295 0.004310329 0.004487269 blood
## Hi_HL60_2 0.996008108 0.003203810 0.005795405 0.005782189 blood
## Hi_HL60_4 0.986202938 0.002948087 0.016482828 0.004618899 blood
## Hi_HL60_6 0.997596728 0.001202118 0.007585141 0.004881700 blood
## Hi_HL60_7 0.982888385 0.006903896 0.009560793 0.005387137 blood
## Hi_HL60_14 0.997717195 0.001131019 0.005303001 0.005319515 blood
## Hi_HL60_15 0.995420209 0.007555687 0.008033794 0.005274234 blood
## Hi_HL60_23 0.994897770 0.002550216 0.006079878 0.005537577 blood
## Hi_HL60_33 0.996792166 0.004219183 0.006055872 0.005153275 blood
## Hi_HL60_36 0.995829430 0.003149186 0.005590083 0.005020308 blood
## Hi_HL60_37 0.996538603 0.006027821 0.008660022 0.005188812 blood
## Hi_HL60_43 0.996048487 0.003662060 0.005691881 0.004947632 blood
## Hi_HL60_48 0.996612696 0.004402476 0.007093712 0.005819998 blood
## Hi_HL60_52 0.995951072 0.004375453 0.007077423 0.004929166 blood
## Hi_HL60_54 0.997391250 0.003718601 0.006168916 0.005145478 blood
## Hi_iPS_1 0.006773016 0.006315371 0.005107275 0.956059574 pluripotent
## Hi_iPS_4 0.005459630 0.005991352 0.006274916 0.938796808 pluripotent
## Hi_iPS_7 0.007263328 0.006166505 0.003852328 0.968530750 pluripotent
## Hi_iPS_8 0.006315538 0.004690963 0.087484043 0.919965537 pluripotent
## Hi_iPS_10 0.005696906 0.006815225 0.002949809 0.962061727 pluripotent
## Hi_iPS_16 0.005472227 0.006513774 0.004396123 0.957432101 pluripotent
## Hi_iPS_23 0.006540182 0.005926688 0.005573124 0.937263083 pluripotent
## Hi_Kera_2 0.005790139 0.990108122 0.007238154 0.004516955 dermal
## Hi_Kera_6 0.007670905 0.986628692 0.006459826 0.005081744 dermal
## Hi_Kera_7 0.003594498 0.996075392 0.006550450 0.003735582 dermal
## Hi_Kera_8 0.006915099 0.996701772 0.007179775 0.003989066 dermal
## Hi_Kera_9 0.010128375 0.979702056 0.005616018 0.004241997 dermal
## Hi_Kera_10 0.002340482 0.996582641 0.006275293 0.004080765 dermal
## Hi_Kera_11 0.004360678 0.992117012 0.006753084 0.003618307 dermal
## Hi_Kera_13 0.006507960 0.997093650 0.006786817 0.003808418 dermal
## Hi_Kera_14 0.013371787 0.984571905 0.006841817 0.004315667 dermal
## Hi_Kera_16 0.012030558 0.970493319 0.005210746 0.004420132 dermal
## Hi_Kera_25 0.004569195 0.994097757 0.006362785 0.003744316 dermal
## Hi_Kera_39 0.004015330 0.990008026 0.006202351 0.003942243 dermal
## Hi_Kera_40 0.077702287 0.758051254 0.006991119 0.004444644 unassigned
## Hi_GW21.2_2 0.005387813 0.005735801 0.997628545 0.004467724 neural
## Hi_GW21.2_7 0.006159599 0.005077643 0.999277999 0.005481231 neural
## Hi_GW21.2_8 0.007246919 0.005188322 0.999270411 0.006086956 neural
## Hi_GW21.2_10 0.006596429 0.004754111 0.999664466 0.006451112 neural
## Hi_GW21.2_13 0.007135911 0.005351180 0.998825375 0.006651760 neural
## Hi_GW21.2_14 0.036515341 0.009006564 0.985959169 0.006464610 neural
## Hi_GW21_3 0.006739270 0.003618348 0.999669848 0.004820364 neural
## Hi_GW21_5 0.010231166 0.013654829 0.998429909 0.003903507 neural
## Hi_NPC_3 0.004529544 0.005334441 0.998956256 0.004840877 neural
## Hi_NPC_10 0.004695227 0.004655696 0.998736001 0.004358722 neural
## Hi_NPC_13 0.005328657 0.004962464 0.993770313 0.006186888 neural
## Hi_GW16_1 0.009173163 0.005942934 0.998591466 0.004733551 neural
## Hi_GW16_2 0.012745847 0.006665138 0.996897389 0.004281073 neural
## Hi_GW16_8 0.007429963 0.004603408 0.994664971 0.005579737 neural
## Hi_GW16_9 0.009558793 0.007997607 0.997377036 0.003998296 neural
## Hi_GW16_20 0.005432284 0.004486833 0.999827641 0.003163685 neural
## Hi_GW16_23 0.024155920 0.004874453 0.999093780 0.004073372 neural
## Hi_GW16_24 0.008210821 0.003561112 0.999554156 0.004285569 neural
## Hi_GW16_26 0.008582737 0.007208059 0.998088158 0.003999380 neuralWe can plot the probabilities and compare our predictions to the true cell types
predictions %>%
mutate(true = test_info$cell_type2) %>%
gather(key = "model", value = "probability", 1:4) %>%
ggplot() +
aes(probability, fill = true) +
geom_histogram(color = "black") +
geom_vline(xintercept = 0.9, color = "red") +
facet_wrap(~model) +
theme_bw() +
scale_fill_viridis_d()## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
In the previous plot, each panel represents the prediction model for each cell type with the output distribution probabilities. The colors in all plot represents the “true”. We can observe that all blood, neural and pluripotent cells were correctly classified using a threshold of 0.9. Only one dermal cell was labeled as unassigned as it was below the threshold. This cell has a probability of 0.75 of being dermal.
Reproducibility
options(width = 70)
devtools::session_info(include_base = TRUE)## Warning: 'DESCRIPTION' file has an 'Encoding' field and re-encoding is
## not possible## Session info --------------------------------------------------------## setting value
## version R version 3.5.0 (2018-04-23)
## system x86_64, darwin15.6.0
## ui X11
## language (EN)
## collate en_AU.UTF-8
## tz Australia/Brisbane
## date 2018-07-24## Packages ------------------------------------------------------------## package * version date
## abind 1.4-5 2016-07-21
## assertthat 0.2.0 2017-04-11
## backports 1.1.2 2017-12-13
## base * 3.5.0 2018-04-24
## bindr 0.1.1 2018-03-13
## bindrcpp * 0.2.2 2018-03-29
## Biobase * 2.40.0 2018-05-01
## BiocGenerics * 0.26.0 2018-05-01
## BiocParallel * 1.14.1 2018-05-06
## bitops 1.0-6 2013-08-17
## broom 0.4.5 2018-07-03
## caret * 6.0-80 2018-05-26
## cellranger 1.1.0 2016-07-27
## class 7.3-14 2015-08-30
## cli 1.0.0 2017-11-05
## codetools 0.2-15 2016-10-05
## colorspace 1.3-2 2016-12-14
## compiler 3.5.0 2018-04-24
## crayon 1.3.4 2017-09-16
## CVST 0.2-2 2018-05-26
## datasets * 3.5.0 2018-04-24
## ddalpha 1.3.4 2018-06-23
## DelayedArray * 0.6.1 2018-06-15
## DEoptimR 1.0-8 2016-11-19
## devtools 1.13.6 2018-06-27
## digest 0.6.15 2018-01-28
## dimRed 0.1.0 2017-05-04
## dplyr * 0.7.6 2018-06-29
## DRR 0.0.3 2018-01-06
## evaluate 0.10.1 2017-06-24
## forcats * 0.3.0 2018-02-19
## foreach 1.4.4 2017-12-12
## foreign 0.8-70 2017-11-28
## GenomeInfoDb * 1.16.0 2018-05-01
## GenomeInfoDbData 1.1.0 2018-05-28
## GenomicRanges * 1.32.3 2018-05-16
## geometry 0.3-6 2015-09-09
## ggExtra * 0.8 2018-04-04
## ggplot2 * 3.0.0 2018-07-03
## glue 1.2.0 2017-10-29
## gower 0.1.2 2017-02-23
## graphics * 3.5.0 2018-04-24
## grDevices * 3.5.0 2018-04-24
## grid 3.5.0 2018-04-24
## gtable 0.2.0 2016-02-26
## haven 1.1.2 2018-06-27
## hms 0.4.2 2018-03-10
## htmltools 0.3.6 2017-04-28
## httpuv 1.4.4.2 2018-07-02
## httr 1.3.1 2017-08-20
## ipred 0.9-6 2017-03-01
## IRanges * 2.14.10 2018-05-16
## irlba * 2.3.2 2018-01-11
## iterators 1.0.9 2017-12-12
## jsonlite 1.5 2017-06-01
## kernlab * 0.9-26 2018-04-30
## knitr 1.20 2018-02-20
## labeling 0.3 2014-08-23
## later 0.7.3 2018-06-08
## lattice * 0.20-35 2017-03-25
## lava 1.6.2 2018-07-02
## lazyeval 0.2.1 2017-10-29
## lubridate 1.7.4 2018-04-11
## magic 1.5-8 2018-01-26
## magrittr * 1.5 2014-11-22
## MASS 7.3-50 2018-04-30
## Matrix * 1.2-14 2018-04-13
## matrixStats * 0.53.1 2018-02-11
## memoise 1.1.0 2017-04-21
## methods * 3.5.0 2018-04-24
## mime 0.5 2016-07-07
## miniUI 0.1.1.1 2018-05-18
## mnormt 1.5-5 2016-10-15
## ModelMetrics 1.1.0 2016-08-26
## modelr 0.1.2 2018-05-11
## munsell 0.5.0 2018-06-12
## nlme 3.1-137 2018-04-07
## nnet 7.3-12 2016-02-02
## parallel * 3.5.0 2018-04-24
## pillar 1.2.3 2018-05-25
## pkgconfig 2.0.1 2017-03-21
## pls 2.6-0 2016-12-18
## plyr 1.8.4 2016-06-08
## pROC * 1.12.1 2018-05-06
## prodlim 2018.04.18 2018-04-18
## promises 1.0.1 2018-04-13
## psych 1.8.4 2018-05-06
## purrr * 0.2.5 2018-05-29
## R6 2.2.2 2017-06-17
## RColorBrewer 1.1-2 2014-12-07
## Rcpp 0.12.17 2018-05-18
## RcppRoll 0.3.0 2018-06-05
## RCurl 1.95-4.10 2018-01-04
## readr * 1.1.1 2017-05-16
## readxl 1.1.0 2018-04-20
## recipes 0.1.3 2018-06-16
## reshape2 1.4.3 2017-12-11
## rlang 0.2.1 2018-05-30
## rmarkdown 1.10 2018-06-11
## robustbase 0.93-1 2018-06-23
## rpart 4.1-13 2018-02-23
## rprojroot 1.3-2 2018-01-03
## rstudioapi 0.7 2017-09-07
## rvest 0.3.2 2016-06-17
## S4Vectors * 0.18.3 2018-06-08
## scales 0.5.0 2017-08-24
## scPred * 0.0.0.9000 2018-07-24
## sfsmisc 1.1-2 2018-03-05
## shiny 1.1.0 2018-05-17
## SingleCellExperiment * 1.2.0 2018-05-01
## splines 3.5.0 2018-04-24
## stats * 3.5.0 2018-04-24
## stats4 * 3.5.0 2018-04-24
## stringi 1.2.3 2018-06-12
## stringr * 1.3.1 2018-05-10
## SummarizedExperiment * 1.10.1 2018-05-11
## survival 2.42-4 2018-06-30
## tibble * 1.4.2 2018-01-22
## tidyr * 0.8.1 2018-05-18
## tidyselect 0.2.4 2018-02-26
## tidyverse * 1.2.1 2017-11-14
## timeDate 3043.102 2018-02-21
## tools 3.5.0 2018-04-24
## utils * 3.5.0 2018-04-24
## viridisLite 0.3.0 2018-02-01
## withr 2.1.2 2018-03-15
## xml2 1.2.0 2018-01-24
## xtable 1.8-2 2016-02-05
## XVector 0.20.0 2018-05-01
## yaml 2.1.19 2018-05-01
## zlibbioc 1.26.0 2018-05-01
## source
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## local
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## Bioconductor
## Bioconductor
## Bioconductor
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## local
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## local
## CRAN (R 3.5.0)
## Bioconductor
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.1)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## Bioconductor
## Bioconductor
## Bioconductor
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## local
## local
## local
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## Bioconductor
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## local
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## local
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## Bioconductor
## CRAN (R 3.5.0)
## local (IMB-Computational-Genomics-Lab/scPred@NA)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## Bioconductor
## local
## local
## local
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## Bioconductor
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## local
## local
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## CRAN (R 3.5.0)
## Bioconductor
## CRAN (R 3.5.0)
## Bioconductor