Select Git revision
train_1_beebook.prob2project
Code owners
Assign users and groups as approvers for specific file changes. Learn more.
sysBiolAlg_lmomaClass.R 15.15 KiB
# sysBiolAlg_lmomaClass.R
# FBA and friends with R.
#
# Copyright (C) 2010-2014 Gabriel Gelius-Dietrich, Dpt. for Bioinformatics,
# Institute for Informatics, Heinrich-Heine-University, Duesseldorf, Germany.
# All right reserved.
# Email: geliudie@uni-duesseldorf.de
#
# This file is part of sybil.
#
# Sybil is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# Sybil is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with sybil. If not, see <http://www.gnu.org/licenses/>.
#------------------------------------------------------------------------------#
# definition of the class sysBiolAlg_lmoma #
#------------------------------------------------------------------------------#
setClass(Class = "sysBiolAlg_lmoma",
contains = "sysBiolAlg"
)
#------------------------------------------------------------------------------#
# default constructor #
#------------------------------------------------------------------------------#
# contructor for class sysBiolAlg_lmoma
setMethod(f = "initialize",
signature = "sysBiolAlg_lmoma",
definition = function(.Object,
model,
wtflux,
COBRAflag = FALSE,
wtobj = NULL,
wtobjLB = TRUE,
obj_coefD = NULL,
absMAX = SYBIL_SETTINGS("MAXIMUM"),
useNames = SYBIL_SETTINGS("USE_NAMES"),
cnames = NULL,
rnames = NULL,
pname = NULL,
scaling = NULL,
writeProbToFileName = NULL, ...) {
if ( ! missing(model) ) {
if (missing(wtflux)) {
tmp <- .generateWT(model, ...)
wtflux <- tmp$fluxes[tmp$fldind]
wtobj <- tmp$obj
}
stopifnot(is(model, "modelorg"),
is(COBRAflag, "logical"),
is(wtobjLB, "logical"),
is(wtflux, "numeric"),
is(absMAX, "numeric"))
stopifnot(length(wtflux) == react_num(model))
# the problem: minimize
#
# | | | ]
# Swt | 0 | 0 | = 0 ] left out if not COBRA
# | | | ]
# -------------------------
# | | |
# 0 | Sdel | 0 | = 0
# | | |
# -------------------------
# | | |
# v1 - v2 |delta-| >= 0
# v2 - v1 |delta+| >= 0
# | | |
# -------------------------
# c_wt | 0 | 0 | >= c^T * v_wt ] left out if not COBRA
# | | |
# lb v_wt |del_lb| 0 |
# ub v_wt |del_ub|10000 |
# | | |
# | | |
# obj 0 | 0 | 1 |
# ---------------------------------------------
# problem dimensions
# ---------------------------------------------
nc <- react_num(model)
nr <- met_num(model)
nCols <- 4*nc
nRows <- ifelse(isTRUE(COBRAflag),
2*nr + 2*nc + 1,
nr + 2*nc)
if (is.null(obj_coefD)) {
deltaobj <- rep(1, 2*nc)
}
else {
stopifnot(length(obj_coefD) == 2*nc)
deltaobj <- obj_coefD
}
# ---------------------------------------------
# constraint matrix
# ---------------------------------------------
# the initial matrix dimensions
LHS <- Matrix::Matrix(0,
nrow = nr + 2*nc,
ncol = nCols,
sparse = TRUE)
# rows for the wild type strain
LHS[1:nr,(nc+1):(2*nc)] <- S(model)
# location of the wild type strain
fi <- c((nc+1):(2*nc))
# rows for the delta match matrix
diag(LHS[(nr+1) :(nr+2*nc),1 :(2*nc)]) <- -1
diag(LHS[(nr+1) :(nr+2*nc),(2*nc+1):(4*nc)]) <- 1
diag(LHS[(nr+1) :(nr+nc) ,(nc+1) :(2*nc)]) <- 1
diag(LHS[(nr+nc+1):(nr+2*nc),1 :nc ]) <- 1
# contraint matrix for linearMOMA, COBRA version if (isTRUE(COBRAflag)) {
# rows for the wild type strain
LHSwt <- Matrix::Matrix(0,
nrow = nr,
ncol = nCols,
sparse = TRUE)
LHSwt[1:nr,1:nc] <- S(model)
# fix the value of the objective function
crow <- Matrix::Matrix(c(obj_coef(model), rep(0, 3*nc)),
nrow = 1,
ncol = nCols,
sparse = TRUE)
# the final contraint matrix
LHS <- rBind(LHSwt, LHS, crow)
subalg <- "lmoma_cobra"
}
else {
subalg <- "lmoma"
}
# ---------------------------------------------
# lower and upper bounds
# ---------------------------------------------
if (isTRUE(COBRAflag)) {
# Here we calculate wild type and deletion strain
# simultaineously, so we need upper and lower bounds
# for both, the wild type and the deletion strain.
# All the delta's are positive.
lower <- c(lowbnd(model),
lowbnd(model),
rep(0, 2*nc))
upper <- c(uppbnd(model),
uppbnd(model),
rep(absMAX, 2*nc))
rlower <- c(rep(0, nRows-1), wtobj)
rupper <- c(rep(0, 2*nr), rep(absMAX, 2*nc), wtobj)
#rupper <- rlower
#rupper <- c(rep(0, 2*nr), rep(0, 2*nc), 0)
}
else {
# Here, we keep the wild type flux distribution fixed.
lower <- c(wtflux, lowbnd(model), rep(0, 2*nc))
upper <- c(wtflux, uppbnd(model), rep(absMAX, 2*nc))
rlower <- c(rep(0, nRows))
rupper <- c(rep(0, nr), rep(absMAX, 2*nc))
}
# ---------------------------------------------
# constraint type
# ---------------------------------------------
if (isTRUE(COBRAflag)) {
rtype <- c(rep("E", 2*nr), rep("L", 2*nc))
if (isTRUE(wtobjLB)) {
rtype <- append(rtype, "L")
}
else {
rtype <- append(rtype, "U")
}
}
else {
rtype <- c(rep("E", nr), rep("L", 2*nc)) }
# ---------------------------------------------
# objective function
# ---------------------------------------------
cobj <- c(rep(0, 2*nc), deltaobj)
# ---------------------------------------------
# row and column names for the problem object
# ---------------------------------------------
if (isTRUE(useNames)) {
if (is.null(cnames)) {
cn <- c(paste("wt", react_id(model), sep = "_"),
paste("del", react_id(model), sep = "_"),
paste("dM", react_id(model), sep = "_"),
paste("dP", react_id(model), sep = "_")
)
colNames <- .makeLPcompatible(cn,
prefix = "x")
}
else {
stopifnot(is(cnames, "character"),
length(cnames) == nCols)
colNames <- cnames
}
if (is.null(rnames)) {
if (isTRUE(COBRAflag)) {
rn <- c(paste("wt", met_id(model), sep = "_"),
paste("del", met_id(model), sep = "_"),
paste("deltaM", 1:nc, sep = "_"),
paste("deltaP", 1:nc, sep = "_"),
"obj_wt"
)
}
else {
rn <- c(paste("del", met_id(model), sep = "_"),
paste("deltaM", 1:nc, sep = "_"),
paste("deltaP", 1:nc, sep = "_")
)
}
rowNames <- .makeLPcompatible(rn,
prefix = "r")
}
else {
stopifnot(is(rnames, "character"),
length(rnames) == nRows)
rowNames <- rnames
}
if (is.null(pname)) {
probName <- .makeLPcompatible(
paste(toupper(subalg), mod_id(model), sep = "_"),
prefix = "")
}
else {
stopifnot(is(pname, "character"),
length(pname) == 1)
probName <- pname
}
}
else {
colNames <- NULL
rowNames <- NULL
probName <- NULL
}
# ---------------------------------------------
# build problem object
# ---------------------------------------------
.Object <- callNextMethod(.Object,
sbalg = subalg,
pType = "lp",
scaling = scaling,
fi = fi,
nCols = nCols,
nRows = nRows,
mat = LHS,
ub = upper,
lb = lower,
obj = cobj,
rlb = rlower,
rub = rupper,
rtype = rtype,
lpdir = "min",
ctype = NULL,
cnames = colNames,
rnames = rowNames,
pname = probName,
algPar = list("wtflux" = wtflux,
"wtobj" = wtobj),
...)
if (!is.null(writeProbToFileName)) {
writeProb(problem(.Object),
fname = as.character(writeProbToFileName))
}
# # ---------------------------------------------
# # build problem object
# # ---------------------------------------------
#
# lp <- optObj(solver = solver, method = method)
# lp <- initProb(lp, nrows = nRows, ncols = nCols)
#
# # ---------------------------------------------
# # set control parameters
# # ---------------------------------------------
#
# if (!any(is.na(solverParm))) {
# setSolverParm(lp, solverParm)
# }
#
#
# loadLPprob(lp,
# nCols = nCols,
# nRows = nRows,
# mat = LHS,
# ub = upper,
# lb = lower,
# obj = cobj,
# rlb = rlower,
# rub = rupper,
# rtype = rtype,
# lpdir = "min"
# )
#
# if (!is.null(scaling)) {
# scaleProb(lp, scaling)
# }
#
# .Object@problem <- lp
# .Object@algorithm <- subalg# .Object@nr <- as.integer(nRows)
# .Object@nc <- as.integer(nCols)
# .Object@fldind <- as.integer(fi)
# validObject(.Object)
}
return(.Object)
}
)
#------------------------------------------------------------------------------#