C13solver

Combination(n, k)

Produces the array of combinations possible picking k from n adapted from Combinadics http://msdn.microsoft.com/en-us/library/aa289166(VS.71).aspx

USAGE:

[out] = Combination (n, k)

INPUTS:

• n – number of elements in the pool

• k – number of elements to pick from n

OUTPUT:

out – array of combinations

cdv2idv(n)

Transformation matrix to transform cumomers to idv’s. idv = cdv2idv(log2(length(cdv)))*cdv; Employs memoization.

USAGE:

[out] = cdv2idv (n)

INPUT:

n – cdv

OUTPUT:

out – idv

errorComputation2(x, Prob)

Computes the total C13 data-fit error (objective value) for a set of fluxes given the experimental data and model carried in Prob

USAGE:

[out] = errorComputation2 (x, Prob)

INPUTS:

• x – flux vector (in null-space / alpha coordinates) at which the error is evaluated

• Prob – problem structure supplied by the solver, with field:

  • .user - structure of user data; .user.expdata (experimental data) and .user.model (model structure) are read here

OUTPUT:

out – scalar C13 data-fit error; when .user.objective is set, the penalised linear objective is returned instead

errorComputation2_grad(x, Prob)

Computes the finite-difference gradient of the C13 data-fit error (errorComputation2) with respect to the fluxes x

USAGE:

[out] = errorComputation2_grad (x, Prob)

INPUTS:

• x – flux vector (in null-space / alpha coordinates) at which the gradient is evaluated

• Prob – problem structure supplied by the solver, with field:

  • .user - structure of user data; .user.model and, when present, .user.diff_interval and .user.useparfor are read here

OUTPUT:

out – gradient vector of the C13 data-fit error, same size as x

generateIsotopomerSolver(model, inputMet, experiment, FVAflag)

Prints a file which looks like BiosyntheticMappingFile except that it has the indexes of every reaction in there as well. After that it calls converter.pl, optimizer.pl and validator.pl but I can take care of that.

USAGE:

generateIsotopomerSolver (model, inputMet, experiment, FVAflag)

INPUTS:

• model – model structure with fields:

  • .isotopomer - cell array of isotopomer mapping strings, one per reaction

  • .rxns - n x 1 cell array of reaction identifiers

• inputMet – input metabolites

• experiment – experiment structure with field:

  • .fragments - structure of measured metabolite fragments, one field per fragment

• FVAflag – default = false, if true then additinoal operations involving fluxVariability involved

Prints a file to /isotopomer/solver/ directory

idv2cdv(n)

Returns transformation to go from idv to cumomers. cdv = idv2cdv(log2(length(idv)))*idv;

USAGE:

[out] = idv2cdv (n)

INPUT:

n – idv

OUTPUT:

out – cdv

idv2idv(n)

Outputs a transformation matrix for changing from forward to reverse order.

order 1 (Jennie’s)

000, 001, 010, 011, 100, 101, 110, 111

order 2 (mine)

000, 100, 010, 110, 001, 101, 011, 111

USAGE:

[out] = idv2idv (n)

INPUT:

n – matrix, size of matrix (2^n x 2^n)

OUTPUT:

out – transforamtion matrix

idv2mdv(n, fragment)

Returns transofmation matrix from idv’s (either Jennie’s or Jan’s order). MDV = idv2mdv(log2(length(idv)))*idv;

USAGE:

[out] = idv2mdv (n, fragment)

INPUT:

n – matrix

OPTIONAL INPUT:

fragment – a vector of carbons to be included. [ 0, 0, 1, 1, 1]’ = last 3 carbons.

OUTPUT:

out – transformation matrix

ratioScore(x, Prob)

Computes a flux-ratio objective (ration’*x)/(ratiod’*x) used when computing confidence intervals on flux ratios

USAGE:

[out] = ratioScore (x, Prob)

INPUTS:

• x – flux vector (in null-space / alpha coordinates)

• Prob – problem structure supplied by the solver, with field:

  • .user - structure of user data; .user.ration (numerator coefficients) and .user.ratiod (denominator coefficients) are read here

OUTPUT:

out – scalar value of the flux ratio at x

ratioScore_grad(x, Prob)

Computes the analytical gradient of the flux-ratio objective evaluated by ratioScore, with respect to the fluxes x

USAGE:

[out] = ratioScore_grad (x, Prob)

INPUTS:

• x – flux vector (in null-space / alpha coordinates)

• Prob – problem structure supplied by the solver, with field:

  • .user - structure of user data; .user.ration (numerator coefficients) and .user.ratiod (denominator coefficients) are read here

OUTPUT:

out – gradient vector of the flux ratio at x, same size as x

scoreC13Fit(flux, expdata, model, namesset, method)

This function (1) computes the theoretical mdv distribution vector for a given flux vector, v, (2) and then computes an error score by taking a running sum of the squared difference between the theortical and experimental mdv vectors.

USAGE:

[output] = scoreC13Fit (flux, expdata, model, namesset, method)

INPUTS:

• flux – flux vector

• expdata – experimental data structure with fields:

  • .std2 - measurement standard deviation used to normalise the error

  • .fragments - structure of measured metabolite fragments, one field per fragment

  • .input - substrate label distribution in cumomer format (used by method 1)

  • .inputfrag - substrate label distribution in EMU/fragment format (used by method 2)

• model – model structure with fields:

  • .lb - n x 1 lower flux bounds (used to test whether flux is a full flux vector)

  • .N - basis of the null space of S, mapping alpha coordinates back to fluxes

• namesset – set of names

• method – method 1 = cumomer, method 2 = CMU

OUTPUT:

output – contains fields:

• error - the calculated error sum value

• theory - theoretical mdv vector

• experimental - experimental mdv vector

Example

v - flux vector array expdata - experimental data structure

e.g.

ala57

• met = xalaL

• fragment = [1,1,1]’

• data = [0.238,0.098,0.017]’

glc_cdv is a sugar distribution in cumomer format (see idv2cdv).

solveLin(A, B)

Solves the linear system A*x = B restricted to the connected block of variables reachable from the non-zero entries of B, leaving the remaining variables at zero

USAGE:

[x] = solveLin (A, B)

INPUTS:

• A – square coefficient matrix of the linear system

• B – right-hand side; a single column is solved directly, while for multiple columns the columns 2:end are solved and the first column is set so that each row of x sums to 1

OUTPUT:

x – solution the same size as B; entries outside the connected block are left at zero