sybilEFBA (version 1.0.2)

FECorr: Function: FECorr: Flux Expression Correlation

Description

This function uses FVA under different conditions to find fluxes that linearly correlates to corresponding gene expression.

Usage

FECorr (model, nCond, initCond, geneExpressionData=NULL,RuleExpressionData=NULL,
				  pct_objective=100,
				  selected_rxns=NULL,
				  only_identified_rules=FALSE,
				  minExprFoldChange=0,
				 lpdir = SYBIL_SETTINGS("OPT_DIRECTION"),
				  solver = SYBIL_SETTINGS("SOLVER"),
			      method = SYBIL_SETTINGS("METHOD"),
				  solverParm=data.frame(CPX_PARAM_EPRHS=1e-6), verboseMode = 2)

Arguments

model

An object of class modelorg.

nCond

Number of conditions (FBA problems to be solved)

geneExpressionData

a data frame: geneID,Cond_id, ExpressionVal column rows are genes and column j+1 is representing gene expression under condition j

RuleExpressionData

rxn_id,cond_id, ExpressionVal

initCond

rxn_id,cond_id,lb,ub,objcoef : lower and upper bounds of rxns under different conditions that represents the available nutrients under these conditions

pct_objective

Biomass will be garnteed to be at least this value multiplied by max biomass calculated using standard FBA. Values are 0 to 100.

selected_rxns

optional parameter used to select a set of reactions not all, Boolean with the same length react_id(model)

only_identified_rules

ignore rxns containing genes with unidentified expression

minExprFoldChange

can be used to consider only genes with a significant change in expression level (i.e min(expression(gene))*minExprFoldChange*2 must be less than or equal to max(expression(gene)))

lpdir

Character value, direction of optimisation. Can be set to "min" or "max". Default: SYBIL_SETTINGS("OPT_DIRECTION").

solver

Single character string giving the solver package to use. See SYBIL_SETTINGS for possible values. Default: SYBIL_SETTINGS("SOLVER").

method

Single character string giving the method the desired solver has to use. SYBIL_SETTINGS for possible values. Default: SYBIL_SETTINGS("METHOD").

solverParm

A named data frame or list containing parameters for the specified solver. Parameters can be set as data frame or list: solverParm = list(parm1 = val1, parm2 = val2) with parm1 and parm2 being the names of two different parameters and val1 and val2 the corresponding values. For possible parameters and values see the documentation of the used solver package (e.g. glpkAPI). Default: SYBIL_SETTINGS("SOLVER_CTRL_PARM").

verboseMode

An integer value indicating the amount of output to stdout: 0: nothing, 1: status messages, 2: like 1 plus with more details, 3: generates files of the LP problem. Default: 2.

Value

returns a list containing slots: geneID,slope,intercept,base_level: OGOR: one gene one rxn iFlux: new fluxes calculated at all the given conditions which considers gene expression data to get a linear fit between gene expression and fluxes. It is a data frame containing the following columns rxn_id: reactionId in model,cond_id:,lb,ub,objCoef,xpc_flux,fva_min,fva_max,RuleExprVal,iflux

Details

Main steps 1- Run FVA for all conditions, exclude rxns fixed in all conditions 2- Identify ruleExpression for set of rxns remaining from 1, 3- Fit Expr to FVA range. 4- Run findMDCFlux to find closest genome-scale flux 5- Recalculate correlation: Posterior, iFlux, ruleExpr

See Also

modelorg, optimizeProb, gene2Rule

Examples

## Not run: 
# 	library(sybil)
# 	data(iAF1260)
# 	model= iAF1260
# 	# trivial test 1, linear levels for an unbounded rxn
# 	ncnd=3
# 	
# 
# 	rxn="R_ADK1"
# 	nc=length(react_id(model))
# 	initCond=cbind(rxn_id=react_id(model),cond_id=rep(1,nc),
# 	         lb=lowbnd(model),ub=uppbnd(model),obj=obj_coef(model))
# 	initCond=rbind(initCond,cbind(rxn_id=react_id(model),
# 	        cond_id=rep(2,nc),lb=lowbnd(model),ub=uppbnd(model),
# 	        obj=obj_coef(model)))
# 	initCond=rbind(initCond,cbind(rxn_id=react_id(model),
# 	     cond_id=rep(3,nc),lb=lowbnd(model),ub=uppbnd(model),
# 		 obj=obj_coef(model)))
# 	
# 	cnds=(1:3)
# 	gprExp=cbind(rxn_id=rxn,cond_id=1,expr_val=2)
# 	gprExp=rbind(gprExp,cbind(rxn_id=rxn,cond_id=2,expr_val=4))
# 	gprExp=rbind(gprExp,cbind(rxn_id=rxn,cond_id=3,expr_val=6))
# 	
# 	fcflx=FECorr(model,nCond=ncnd,initCond=initCond,
# 	       RuleExpressionData=gprExp,selected_rxns=(react_id(model)==rxn),
# 		   verboseMode=4);
# 	fcflx[[2]][fcflx[[2]][,2]==rxn,]
# 	
# 	
# 	cor(as.numeric(fcflx[[2]][fcflx[[2]][,2]==rxn,"expr_val"]),
# 	               as.numeric(fcflx[[2]][fcflx[[2]][,2]==rxn,"iflx"]))
# 
# 
# ## End(Not run)