| Title: | Inferring unobserved perturbations from gene expression data |
|---|---|
| Description: | Takes as input an incomplete perturbation profile and differential gene expression in log odds and infers unobserved perturbations and augments observed ones. The inference is done by iteratively inferring a network from the perturbations and inferring perturbations from the network. The network inference is done by Nested Effects Models. |
| Authors: | Martin Pirkl [aut, cre] |
| Maintainer: | Martin Pirkl <[email protected]> |
| License: | GPL-3 |
| Version: | 1.13.0 |
| Built: | 2024-09-18 04:56:26 UTC |
| Source: | https://github.com/bioc/nempi |
Builds and uses different classifiers to infer perturbation profiles
classpi( D, unknown = "", full = TRUE, method = "svm", size = NULL, MaxNWts = 10000, ... )classpi( D, unknown = "", full = TRUE, method = "svm", size = NULL, MaxNWts = 10000, ... )
D |
either a binary effects matrix or log odds matrix as for Nested Effects Models (see package 'nem') |
unknown |
colname of samples without mutation data, E.g. "" |
full |
if FALSE, does not change the known profiles |
method |
either one of svm, nn, rf |
size |
parameter for neural network (see package 'nnet') |
MaxNWts |
parameters for neural network (see package 'nnet') |
... |
additional parameters for mnem::nem |
plot
Martin Pirkl
D <- matrix(rnorm(1000*100), 1000, 100) colnames(D) <- sample(seq_len(5), 100, replace = TRUE) Gamma <- matrix(sample(c(0,1), 5*100, replace = TRUE, p = c(0.9, 0.1)), 5, 100) Gamma <- apply(Gamma, 2, function(x) return(x/sum(x))) Gamma[is.na(Gamma)] <- 0 rownames(Gamma) <- seq_len(5) result <- classpi(D)D <- matrix(rnorm(1000*100), 1000, 100) colnames(D) <- sample(seq_len(5), 100, replace = TRUE) Gamma <- matrix(sample(c(0,1), 5*100, replace = TRUE, p = c(0.9, 0.1)), 5, 100) Gamma <- apply(Gamma, 2, function(x) return(x/sum(x))) Gamma[is.na(Gamma)] <- 0 rownames(Gamma) <- seq_len(5) result <- classpi(D)
Infers perturbations profiles based on a sparse perturbation matrix and differential gene expression as log odds
nempi( D, unknown = "", Gamma = NULL, type = "null", full = TRUE, verbose = FALSE, logtype = 2, null = TRUE, soft = TRUE, combi = 1, converged = 0.1, complete = TRUE, mw = NULL, max_iter = 100, keepphi = TRUE, start = NULL, phi = NULL, ... )nempi( D, unknown = "", Gamma = NULL, type = "null", full = TRUE, verbose = FALSE, logtype = 2, null = TRUE, soft = TRUE, combi = 1, converged = 0.1, complete = TRUE, mw = NULL, max_iter = 100, keepphi = TRUE, start = NULL, phi = NULL, ... )
D |
either a binary effects matrix or log odds matrix as for Nested Effects Models (see package 'nem') |
unknown |
colname of samples without mutation data, E.g. "" |
Gamma |
matrix with expectations of perturbations, e.g. if you have a binary mutation matrix, just normalize the columns to have sum 1 |
type |
"null": does not use the unknown samples for inference at the start, "random" uses them in a random fashion (not recommended) |
full |
if FALSE, does not change the known profiles |
verbose |
if TRUE gives more output during inference |
logtype |
log type for the log odds |
null |
if FALSE does not use a NULL node for uninformative samples |
soft |
if FALSE discretizes Gamma during the inference |
combi |
if combi > 1, uses a more complex algorithm to infer combinatorial perturbations (experimental) |
converged |
the absolute difference of log likelihood till convergence |
complete |
if TRUE uses the complete-data logliklihood (recommended for many E-genes) |
mw |
if NULL infers mixture weights, otherwise keeps them fixed |
max_iter |
maximum iterations of the EM algorithm |
keepphi |
if TRUE, uses the previous phi for the next inference, if FALSE always starts with start network (and empty and full) |
start |
starting network as adjacency matrix |
phi |
if not NULL uses only this phi and does not infer a new one |
... |
additional parameters for the nem function (see package mnem, function nem or mnem::nem) |
nempi object
Martin Pirkl
D <- matrix(rnorm(1000*100), 1000, 100) colnames(D) <- sample(seq_len(5), 100, replace = TRUE) Gamma <- matrix(sample(c(0,1), 5*100, replace = TRUE, p = c(0.9, 0.1)), 5, 100) Gamma <- apply(Gamma, 2, function(x) return(x/sum(x))) Gamma[is.na(Gamma)] <- 0 rownames(Gamma) <- seq_len(5) result <- nempi(D, Gamma = Gamma)D <- matrix(rnorm(1000*100), 1000, 100) colnames(D) <- sample(seq_len(5), 100, replace = TRUE) Gamma <- matrix(sample(c(0,1), 5*100, replace = TRUE, p = c(0.9, 0.1)), 5, 100) Gamma <- apply(Gamma, 2, function(x) return(x/sum(x))) Gamma[is.na(Gamma)] <- 0 rownames(Gamma) <- seq_len(5) result <- nempi(D, Gamma = Gamma)
Bootstrap algorithm to get a more stable result.
nempibs(D, bsruns = 100, bssize = 0.5, replace = TRUE, ...)nempibs(D, bsruns = 100, bssize = 0.5, replace = TRUE, ...)
D |
either a binary effects matrix or log odds matrix as |
bsruns |
number of bootstraps |
bssize |
number of E-genes for each boostrap |
replace |
if TRUE, actual bootstrap, if False sub-sampling |
... |
additional parameters for the function nempi |
list with aggregate Gamma and aggregate causal network phi
Martin Pirkl
D <- matrix(rnorm(1000*100), 1000, 100) colnames(D) <- sample(seq_len(5), 100, replace = TRUE) Gamma <- matrix(sample(c(0,1), 5*100, replace = TRUE, p = c(0.9, 0.1)), 5, 100) Gamma <- apply(Gamma, 2, function(x) return(x/sum(x))) Gamma[is.na(Gamma)] <- 0 rownames(Gamma) <- seq_len(5) result <- nempibs(D, bsruns = 3, Gamma = Gamma)D <- matrix(rnorm(1000*100), 1000, 100) colnames(D) <- sample(seq_len(5), 100, replace = TRUE) Gamma <- matrix(sample(c(0,1), 5*100, replace = TRUE, p = c(0.9, 0.1)), 5, 100) Gamma <- apply(Gamma, 2, function(x) return(x/sum(x))) Gamma[is.na(Gamma)] <- 0 rownames(Gamma) <- seq_len(5) result <- nempibs(D, bsruns = 3, Gamma = Gamma)
Compares the ground truth of a perturbation profile with the inferred profile
pifit(x, y, D, unknown = "", balanced = FALSE, propagate = TRUE, knowns = NULL)pifit(x, y, D, unknown = "", balanced = FALSE, propagate = TRUE, knowns = NULL)
x |
object of class nempi |
y |
object of class mnemsim |
D |
data matrix |
unknown |
label for the unlabelled samples |
balanced |
if TRUE, computes balanced accuracy |
propagate |
if TRUE, propagates the perturbation through the network |
knowns |
subset of P-genes that are known to be perturbed (the other are neglegted) |
list of different accuracy measures: true/false positives/negatives, correlation, area under the precision recall curve, (balanced) accuracy
Martin Pirkl
library(mnem) seed <- 42 Pgenes <- 10 Egenes <- 10 samples <- 100 uninform <- floor((Pgenes*Egenes)*0.1) Nems <- mw <- 1 noise <- 1 multi <- c(0.2, 0.1) set.seed(seed) simmini <- simData(Sgenes = Pgenes, Egenes = Egenes, Nems = Nems, mw = mw, nCells = samples, uninform = uninform, multi = multi, badCells = floor(samples*0.1)) data <- simmini$data ones <- which(data == 1) zeros <- which(data == 0) data[ones] <- rnorm(length(ones), 1, noise) data[zeros] <- rnorm(length(zeros), -1, noise) lost <- sample(1:ncol(data), floor(ncol(data)*0.5)) colnames(data)[lost] <- "" res <- nempi(data) fit <- pifit(res, simmini, data)library(mnem) seed <- 42 Pgenes <- 10 Egenes <- 10 samples <- 100 uninform <- floor((Pgenes*Egenes)*0.1) Nems <- mw <- 1 noise <- 1 multi <- c(0.2, 0.1) set.seed(seed) simmini <- simData(Sgenes = Pgenes, Egenes = Egenes, Nems = Nems, mw = mw, nCells = samples, uninform = uninform, multi = multi, badCells = floor(samples*0.1)) data <- simmini$data ones <- which(data == 1) zeros <- which(data == 0) data[ones] <- rnorm(length(ones), 1, noise) data[zeros] <- rnorm(length(zeros), -1, noise) lost <- sample(1:ncol(data), floor(ncol(data)*0.5)) colnames(data)[lost] <- "" res <- nempi(data) fit <- pifit(res, simmini, data)
Plot function for an object of class 'nempi'.
## S3 method for class 'nempi' plot(x, barlist = list(), heatlist = list(), ...)## S3 method for class 'nempi' plot(x, barlist = list(), heatlist = list(), ...)
x |
object of class 'nempi' |
barlist |
additional arguments for function 'barplot' from package 'graphics' |
heatlist |
additional arguments for function 'HeatmapOP' from package 'epiNEM' |
... |
additional arguments for function 'plotDnf' from package 'mnem' |
Plots of the optimal network phi and perturbation matrix.
Martin Pirkl
D <- matrix(rnorm(1000*100), 1000, 100) colnames(D) <- sample(seq_len(5), 100, replace = TRUE) result <- nempi(D) plot(result)D <- matrix(rnorm(1000*100), 1000, 100) colnames(D) <- sample(seq_len(5), 100, replace = TRUE) result <- nempi(D) plot(result)
Produces different convergence plots based on a nempi object
## S3 method for class 'nempi' plotConvergence(x, type = "b", ...)## S3 method for class 'nempi' plotConvergence(x, type = "b", ...)
x |
nempi object |
type |
see ?plot.default |
... |
additional parameters for plot |
plot
Martin Pirkl
D <- matrix(rnorm(1000*100), 1000, 100) colnames(D) <- sample(seq_len(5), 100, replace = TRUE) Gamma <- matrix(sample(c(0,1), 5*100, replace = TRUE, p = c(0.9, 0.1)), 5, 100) Gamma <- apply(Gamma, 2, function(x) return(x/sum(x))) Gamma[is.na(Gamma)] <- 0 rownames(Gamma) <- seq_len(5) result <- nempi(D, Gamma = Gamma) par(mfrow=c(2,3)) plotConvergence(result)D <- matrix(rnorm(1000*100), 1000, 100) colnames(D) <- sample(seq_len(5), 100, replace = TRUE) Gamma <- matrix(sample(c(0,1), 5*100, replace = TRUE, p = c(0.9, 0.1)), 5, 100) Gamma <- apply(Gamma, 2, function(x) return(x/sum(x))) Gamma[is.na(Gamma)] <- 0 rownames(Gamma) <- seq_len(5) result <- nempi(D, Gamma = Gamma) par(mfrow=c(2,3)) plotConvergence(result)