Change Log for influx_si¶

2024-02-29 version 7.0.2¶

New feature:

influx_si can be called from python as ‘influx_s.main(argv)’
an absence of labeling measurements is now signaled as error in influx_si (issue #6 submitted by Loic Le Gregam, TBI, France)
ftbl2netan get an option ‘–eprl’ for parallel experiments
documentation is excluded from package, see it on https://influx-si.readthedocs.io/

Bug fix:

fixed plot_ilab.R for empty simulated data (seen in case submitted by Cécilia Berges, TBI, France)
fixed python exception for ‘-o outdir’
fixed some warnings written in .err or stdout instead of .log
fixed option –nocalc not working in case of one network
fixed parsing present but empty MTF files
fixed moving .ftbl and .ikib files to tmp/ dir from parallel experiments
fixed typo in ‘.cnstr’ suffix for .vmtf case (issue #7 reported by Rochelle Kouakou, TBI, France)
fixed moving .R files to tmp/ dir in case of multiple inputs when one or several of them finished with an error.

2023-10-04 version 7.0.1¶

Bug fix:

fixed move of .ikin file to tmp/ directory
fixed moving files on Windows
fixed panda warning about concatenation of empty DataFrame

2023-07-11 version 7.0¶

New feature:

new output format is introduced. All result files go in <mynetwork>_res directory
ff2ftbl and res2ftbl_meas are now deprecated. Use *.sim files from new output format.
in influx_i, unscaled simulated mass measurements (in .miso.sim) are clamped to [0; 1] and full MID sets are normalized to sum up to 1 (requested by Noémi Butin and Pierre Millard, TBI, France)
default MTF values lacking in input files are now explicitly written in newly created files .def
scaling factors are now calculated even if –noopt is set.
pooled measurements with spaces around “+” sign are now allowed (seen at Magalie Poirier Pocovi, U. of Bordeaux, France)
plot_imes.R is deprecated, use plot_ilab.R instead
txt2ftbl: now, peak measurements are translated to PEAK_MEASUREMENTS and no more to LABEL_MEASUREMENTS in ftbl

Bug fix:

fixed error messages from funlab in which reported time point could be erroneous.
fixed plot_ilab.R in simulation mode
fixed non ordered time points in .ikin while converted from .miso (seen at Magalie Poirier Pocovi, U. of Bordeaux, France)
txt2ftbl: fixed a possible appearance of non labeled metabolites in METABILITE_POOLS
influx_si: better Exception handling and error reporting
ftbl2labcin: fixed mapping ftbl<->miso for peak and label measurements
influx_si: better handling of file names with a dot “.” in the name

2022-10-11 version 6.1¶

New feature:

species not carrying labeled atoms cannot appear in FTBL/METABOLITE_POOLS
ftbl2cumoAb and other ftbl2… utilities can use ‘–prefix’/’–mtf’ options
ftbl2cumoAb eliminates linearly dependent rows from Afl
ftbl2cumoAb gets ‘–flux’ option to read flux values from a 2-columns tsv file
kvh, nlsic modules are moved to independent package
error about dependent stoichiometric equations is turned into warning

Bug fix:

stoichiometric coefficients in .netw can be written with “*”, not necessarily with “ * “
fixed biomass-like reactions in C13ftbl
made explicit error message when cumomer system has b=0

2022-06-23 version 6.0.4¶

New feature:

Time values <= 0 in .miso are silently filtered out in instationary mode
ftbl2mtf and txt2ftbl now write their results in UTF-8 encoding

Bug fix:

fixed reading MTF and FTBL files not in UTF-8 encoding

2022-06-13 version 6.0.1¶

Bug fix:

fixed inclusion of test/* files, namely mtf/* files (reported by Younes Dellero, INRAE, Rennes, France)

2022-06-08 version 6.0¶

New features:

introduced new front-end MTF format (cf. ftbl2mtf, txt2ftbl)
added plot_ilab.R for plotting instationary MS/NMR measurements and simulations (replaces plot_imass.R)
ftbl2xgmml: added orange color for dead-end metabolites
txt2ftbl: added detection of disconnected sub-networks
ftbl2netan: dead-ends are reported
ftbl2metxml: added pathway information and compartment treatment

Bug fix:

fixed MS measurements on long molecules in –emu mode

2021-02-26 version 5.4.0¶

New features:

influx_i: user can add custom R expressions for label input varying in time (cf. funlabR)
influx_i: explicit error message on non valid entries in label kinetic data (requested by Pierre Millard, INRAE, France)
added optctrl:mumps:icntl_NN option (case reported by Albert Fina Romero, UAB, Spain)
influx_i: comment string in kinetic data has changed to // (# will still work with old files)
added -i option to ftbl2netan (for influx_i-like analysis)

Bug fix:

fixed error raising on redundant equality when --ffguess is used (reported by Pierre Millard, INRAE, France)
fixed fatal error with scale factor infrastructure in some cases when --ffguess is used
plot_imass.R is fixed for a case when there is no any MS measurement
explicit error message on fail while reading label kinetic data
fixed not working error messages in few utilities
fixed too early closed kvh file connection
fixed post-production calculations when optimization in influx_si failed

2020-07-24 version 5.3.0¶

New features:

added PSO optimization method (suggested by Pierre Millard, INRAE, Toulouse, France)
now optimization methods can be chained in the same run, e.g. --meth pso --meth nlsic
the field optctrl... is now formatted to allow homonym parameters with different values for different optimization methods
option --install_rdep becomes interactive and reproduces the behavior of R’s install.packages() to allow installation in a user’s R personal library when the default directory is not writable (suggested by Matthieu Guionnet, CNRS, Toulouse, France)

Bug fix:

plot_smeas.R is fixed for cases where NA is present in MS measurements.

2020-05-28 version 5.2.0¶

New features:

ff2ftbl can accept an input kvh file not ending with _res.kvh
ftbl2netan will write a partial output in case of error during FTBL parsing
added error message if input metabolite appears in measurment FTBL sections
in kvh module added optional boolean parameter ‘strip’

Bug fix:

fixed txt2ftbl, removed xrange() hidden call
fixed kvh.dict2kvh() for writing objects derived from dict()
fixed putting border line starting values inside feasible domain for BFGS method
fixed endline problem in ftbl2metxml and res2ftbl_meas on Windows (reported by Loic Le Gregam, INSA, Toulouse, France)

2020-04-07 version 5.1.0¶

New features:

option –cupn is now considered in absolute sens, i.e. net fluxes are limited from both sides: ‘-cupn <= netflux <= cupn’ (before it was ‘netflux <= cupn’ only)
influx_i.py is no more symbolic link neither just a copy of influx_s.py, just a python proxy based on exec().
e_coli.ftbl file from test collection is configured to create e_coli.pdf with graphics instead of e_coli.RData

Bug fix:

fixed executable files of additional tools
fixed plot_smeas.R for plotting simulated but not measured MS data
fixed plot_smeas.R for plotting repeatedly measured fragments by MS
fixed non reproducible R code generation (Python must be >= 3.6 for this)
fixed kvh reading which was impacting several additional tools
parallel experiment examples are back in the test/prl_exp/ directory

2020-03-05 version 5.0.3¶

New feature:

all Python scripts are doubled with executable files without ‘.py’ extension (suggested by Pierre Millard, INRAE, and Matthieu Guionnet, CNRS, Toulouse, France)

Bug fix:

fixed few typos in documentation

2020-02-26 version 5.0.2¶

Bug fix:

fixed parallel execution of multiple FTBL on windows
fixed influx_si hanging on in-existing directory in FTBL name
fixed dimnames()[[1]]<- failure when nrow==1 (reported by Younes Dellero, INRAE, Rennes, France)
fixed –ffguess silently failing when redundant reaction was present (idem)
fixed use of influx_s.py –install_rdep (reported by Loic Le Gregam, INSA, Toulouse, France)
updated user manual for trouble-shooting and consulting services

2020-01-10 version 5.0.1¶

Bug fix:

fixed fatal error on windows platform
fixed startup message about Rcpp_Rmumps module on some platforms
fixed dependency on libsbml

2019-10-25 version 5.0¶

New features:

converted to Python 3
packaged with python distutils and Conda package management system. Now installation can be made as simple as pip install influx_si or conda install influx_si. Internet connection is required for both methods.
auxiliary C++ routines are placed in multbxxc R packages. So, no more compilation is needed at first execution of influx_si.

Bug fix (all bugs in this release are reported by Baudoin Delépine, INSA, Toulouse, France):

fixed doc about R version (now 3.4.0 or higher) and rmumps (now 5.2.1-3 or higher)
fixed use of –sln option
added explicit error message if tmax is Inf (can happen if time grid could not be read from file indicated in file_labcin ftbl field)

2017-07-04 version 4.4.3¶

New features:

added 95% quantile in monte-carlo/cost/ci field in _res.kvh file. It makes possible to do a mono-tail chi2 test for goodness of fit
added possibility for FTBL files to be encoded in UTF16 and UTF32 (based on case reported by Lucille Stuani, INSERM, Toulouse, France)
added an explicit error message if no label information could be found in parallel experiment FTBL (reported idem)

Bug fix:

fixed a warning in Monte-Carlo iteration about multiple values in if() close (reported idem)

2017-06-15 version 4.4.2¶

New features:

added a new field “constrained net-xch01 fluxes” to the result kvh file
ff2ftbl.py: instead of only free fluxes, all fluxes are read in kvh file. Thus in a modified FTBL, a partition on free/dependent/constrained fluxes can differ from those used in the kvh file.
ff2ftbl.py: if kvh and ftbl files have the same prefix, only this prefix can be given as a unique command line argument

Bug fix:

ff2ftbl.py: fixed “end of line” bug on Windows platform
plot_smeas.R: fixed metabolite names retrieving in parallel experiments
plot_smeas.R: fixed disgraceful exit if simulation in influx_s failed

2017-05-24 version 4.4.1¶

New features:

in plot_smeas.R and plot_imass.R few cosmetic improvements in plot titles and legends

Bug fix:

fixed non varying free pools in influx_s
fixed cases where some cumomer (or EMU) weights can have no cumomers (EMU)
fixed libs.R by including some files (impacted preamble.R)

2017-05-22 version 4.4¶

New features:

added plot_smeas.R file to be included in posttreat_R field in FTBL/OPTIONS. It plots all stationary measurements vs their simulated counterparts in a pdf file.
added preamble.R, an example of starting session when working with mynetwork.RData issued from save_all.R or save_minenv.R
R can be again of version 3.3+ (not necessarily 3.4+)
minor speedup in instationary simulations

Bug fix:

fixed names in dev vector of pool measurements

2017-04-28 version 4.3¶

New features:

speed up of about 30-40% is achieved for instationary simulations with 2nd order time scheme (need upgrade R at least to 3.4.0).
in plot_imass.R, each measured mass fragment is presented in a separate plot instead of regrouping all fragments for a given metabolite on the same plot.
in plot_imass.R, non measured metabolites are plotted too now

Bug fix:

fixed Monte-Carlo iterations with –np=1
added a mention of python-libsbml in installation procedure

2017-03-30 version 4.2¶

New features:

added a script ftbl2metxml.py converting an ftbl to an xml file suitable for visualization on http://metexplore.toulouse.inra.fr . Additionally, it reads flux values from corresponding …_res.kvh file (if available) and put them in files …_net.txt, …_fwd.txt and …_rev.txt for later copy/pasting on the MetExplore site (suggested by Tony Palama, INSA, Toulouse, France)
added comment tags ‘###’ to txt network format (recognized by txt2ftbl.py and respectively ‘//##’ tag in FTBL format recognized by all programs using FTBL format) to mark a new pathway. It allows ftbl2metxml.py to assign reactions to pathways and thus make a network graph more readable.

Bug fix:

a duly error message is added to signal a network without any label entry in a reduced cumomer network.

2017-03-03 version 4.1¶

New features:

added explicit error message when label transitions are missing for any reaction in NETWORK section
improved speed of labeling simulation in influx_i
added parameter –tblimit[=0] for trace back limit in errors generated by python (for developers only).

Bug fix:

fixed error appearing in influx_i during parallel experiments in situation where time intervals are different in different experiments (reported by Maria Fatarova, INSA, Toulouse, France)
fixed file creation in plot_imass.R (it created pdf in the current directory instead of the working one)

2016-12-20 version 4.0.1¶

Bug fix:

file txt2ftbl.py was lacking in the previous version (reported by Tony Palama, INSA, Toulouse, France)

2016-12-19 version 4.0¶

New features:

parallel experiments (i.e. same metabolic state but different label entries) can now be processed both in stationary (influx_s) and instationary (influx_i) labeling
reaction having more than 2 metabolites on ever side can now be entered in FTBL as a series of reactions with the same name
metabolites with no carbon transitions (like ATP, NADP etc. when they are just co-factors) can now be entered in NETWORK section. They can have stoechiometric coefficients different from 1
the same metabolite can now appear on both sides of a reaction. It can be helpful for some special carbon shuffles
reactions without carbon transitions can now be entered in a new FTBL section NOTRACER_NETWORK. It is a good place to enter for example biomass reaction. Stoechiometric coefficients different from one are allowed at this place
new utility txt2ftbl.py translates an easier readable/writable format for chemical reactions to an FTBL file
added –addnoise option to facilitate creation of realistic simulated measurements

Bug fixes:

fixed a bug preventing Monte-Carlo simulations with influx_i (reported by Maria Fatarova, INSA, Toulouse, France)

2016-07-29 version 3.2¶

New features:

added controls for coherence of label transitions
added detection of incoherent fragments in MASS_MEASUREMENTS (e.g. longer one than a whole molecule)
in LABEL_INPUT section, if incomplete labeled forms don’t sum up to 1, and several labeled forms are absent, the lacking label fraction is assigned to the fully unlabeled form
R package ‘snow’ is no more needed on windows platform to run Monte-Carlo simulations in parallel mode
on all platforms, Monte-Carlo simulations are now run on a PSOCK cluster and no more on a FORK cluster (Linux) or SOCK (Windows)
inequalities involving only constrained fluxes or depending solely on such fluxes are now simply ignored with a warning
fixed Jacobian calculation when no free flux exists

Bug fixes:

fixed building a library mult_bxxc.dll on Windows platform (reported by Tony Palama, INSA/MetaToul Toulouse, France)
fixed building mult_bxxc.so in parallel context (multiple ftbls)
fixed formulas in equalities and inequalities with flux names having parenthesis, brackets, spaces and alike
cluster workers are parsimoniously created in case of multiple starting points

2016-06-13 version 3.1¶

New features:

added controls and explicit error messages for DEVIATION=0 in FTBL file
added column “p-value” to residual values in _res.kvh file (may help for outlier choice)
added check-points for infinite values that can appear in residual and Jacobian

Bug fixes:

fixed EMU mode in instationary case
fixed renewal of mult_bxxc.so library in case of source update
fixed cost value calculation in case of outlier exclusion

2016-04-18 version 3.0.1¶

Bug fixes:

fixed including mult_bxxc.so file in zip archive which prevented from proper compiling of this dynamic library
fixed an absence of C++11 flag on platforms where it is not a standard by default.

2016-04-15 version 3.0¶

New features:

influx_i.py is introduced for instationary label modeling
some critical calculations are written in C++ so some compilation is needed at first execution.
new optional package limSolve is used and need to be installed (as well as its dependencies) if --lim option is used
more TIMEIT points introduced for finer time control

Bug fixes:

fixed performance issue in slam package for ‘-’ and ‘+’ operations
fixed sparse matrix preparation when there is only one non zero entry

2016-02-18 version 2.15¶

New features:

calculation speed was improved due to the use of packages slam and rmumps instead of Matrix;
added “cpu” field when timing is requested

Bug fixes:

fixed a bug preventing a use of a flux added in EQALITIES in measured fluxes (reported by Edern Cahoreau, INSA, Toulouse, France)
fixed minor problems in MC iterations (parameter distribution was not significantly affected)

2015-01-19 version 2.14¶

New features:

commandArgs field in FTBL file can have comments in it and occur more than once somewhere in the OPTIONS section
--DEBUG option is removed as obsolete
R package bitops is not required anymore to be installed (valid for R-3.0.0 or higher)

Bug fixes:

fixed a bug in delivering an error message when commandArgs had a comment in it
fixed the precedence of command line options over commandArgs given in FTBL
fixed a bug in parsing FTBL file having a BOM (invisible utf8 encoding mark) in it (reported by Yanfen Fu, University of Washington, USA)
fixed representation of growth fluxes by ftbl2xgmml.py utility

2014-09-17 version 2.13¶

New features:

posttreat_R field can have several file names separated by '; '
added explicit error message if a valid float value is missing for free or constrained flux
added explicit error message if no dependent flux is included in the balance on any metabolite (suggested by a case submitted by Marc O. Warmoes, Cornell University, USA)
in the documentation, added a paragraph about consulting offer
result .kvh file is greatly shortened, keeping only essential information. Custom additional information can be stored in some file via posttreat_R option
now, influx_s returns a non zero code to shell if an error occurred during execution;
added a parameter monotone to the control list of NLSIC.

Bug fixes:

fixed a bug in generating EMU systems (manifested in some special cases)
fixed an error preventing from producing a message suggesting a new partition among dependent, free and constrained fluxes (reported by Stéphane Mottelet, University of Compiegne, France)
fixed metabolite pooling weights (manifested in some special cases)
Windows platform: fixed passing command line options to R code
Windows platform: precompiled nnls R package (version 32 bits) can produce wrong results. Recompile it by hand or use 64 bits version.

2014-07-02 version 2.12¶

New features:

parsing badly formatted ftbl files is made more robust

Bug fixes:

fixed a bug in --emu option (was introduced in v.2.11)

2014-06-12 version 2.11.1¶

Bug fix:

an option --noopt broken in 2.11 is repaired (reported by Pierre Millard, Manchester Institute of Biotechnology, UK)

2014-06-11 version 2.11¶

New features:

a joint use of the options --fseries and --irand gives a possibility to mix fixed and random values in starting points
post treatment option posttreat_R is introduced in FTBL file. A user script written in R can be used to chain flux estimation and customized data treatment, e.g. graph plotting in a pdf file or simply saving of all the environment for later use and exploring in an R interactive session
added optional INEQUALITIES/METAB section in FTBL file. It can be helpful to limit variations of estimated metabolite concentrations (suggested by Marc Carnicer, INSA of Toulouse, France)
added optional EQUALITIES/METAB section in FTBL file. It can be helpful to fix a ratio between varying metabolite concentrations (suggested by Marc Carnicer, idem)
the default value of btdesc parameter in NLSIC algorithm is lowered from 0.75 to 0.1. In some cases, it can accelerate the optimization convergence.

Bug fixes:

fixed EMU list of participants in measurements
fixed measurement matrix when only one measurement is available
fixed a fatal error when no free flux is available but at least one metabolite quantity must be estimated
fixed a bug in pooled measurements. This bug was harmful only if the metabolite pooling was used in more than one type of measurements, e.g. mass and labeling. If only one type of measurements used pooling (e.g. mass), the bug was without effect
where appropriate, a word “labeling” was replaced by “label” in the field names of the _res.kvh file
fixed superfluous backtracking iterations present for some particular residual functions
if a flux or a metabolite is present more than once in formulas of (IN)EQUALITIES sections, its coefficients are summed up instead of taking only the last one
fixed a fatal error in generating inequality matrix for net fluxes

2014-04-08 version 2.10¶

New features:

added an option --tikhreg which is an alternative for --ln option. In case of rank deficient Jacobian, it calculates an increment step of the smallest norm in approximative way. It is done by Tikhonov regularization
added an option --ffguess which makes to ignore the partition between free and dependent fluxes defined in FTBL file(s) and automatically guess a new free/dependent flux partition (suggested by Roland Nilsson, Karolinska Institutet, Sweden)
added utility ftbl2kvh.py which is useful for debugging purposes only
utilities ftbl2xgmml.py, ftbl2cumoAb.py, ftbl2netan.py and ftbl2kvh.py are rewritten in such a way that if no output redirection (with operands ‘>’ or ‘|’) occurs on the command line, the name of the output file is automatically derived from the input one. The suffix .ftbl is simply replaced with .xgmml, .sys, .netan or .kvh respectively. Thus a plain drug-and-drop can work with these utilities
option --TIMEIT reports times with subsecond precision. The actual precision depends on the platform but typically a 0.01 s precision should be available. On Windows, the precision is usually 1/60 of a second

Bug fixes:

fixed include_growth_flux option for ftbl2cumoAb.py utility (reported by Marc Carnicer, INSA of Toulouse, France)
fixed a bug preventing from checking for a linear dependence between rows of stoechiometric matrix if no constrained net flux is defined in the FTBL file (reported by Roland Nilsson, idem)

2014-02-05 version 2.9¶

New features:

utility ftbl2xgmml.py replaces ftbl2rsif.py. Now, a standalone XGMML file describes both a network and its graphical properties instead of a collection of files where this information was spread. New graphical conventions are now used.
an obsolete utility ftbl2cytoscape.bat is removed form the distribution.
added utility res2ftbl_meas.py generating measurement section from a result file _res.kvh
added utility expa2ftbl.R transforming stoechiometric information in EXPA format (http://gcrg.ucsd.edu/Downloads/ExtremePathwayAnalysis) to various sections of FTBL file, namely to EQUALITY section where non carbon carrying fluxes can appear
files generated by influx_s and collecting values for graphical representation (like edge.netflux.mynetwork and others) are renamed by adding a suffix .attrs to make them compatible with Cytoscape v3.0
utilities ffres2ftbl.sh and ff2ftbl.py distributed for a long time ago, are now mentioned in the documentation

Bug fixes:

fixed --fullsys option broken in the previous release.

2014-01-27 version 2.8¶

New features:

EQUALITY section in FTBL file may include fluxes absent in NETWORK section, e.g. fluxes involved in non carbon carrying reactions (suggested by Roland Nilsson, Karolinska Institutet, Sweden)
when a meaningful partition between free and dependent fluxes cannot be made, a proposition is made as to stoechiometric equations to be eliminated by hand (suggested by Roland Nilsson, idem)
when --clownr option is used, reduced size of cumomer system is more efficient than without this option (replace a fix in 2.6 version)

Bug fixes:

fixed useless memory consumption during ftbl parsing when --emu option is used and very long molecules (say >20 carbons) are present (reported by Roland Nilsson, idem)
some error messages are made more explicit during FTBL parsing
fixed Jacobian calculation for condensing input reaction
fixed matrix constructions when no free flux is defined
fixed b term for full cumomer system
fixed inequality enforcement when adaptive backtracking is used in NLSIC
fixed inequality precedence, now specific inequalities from FTBL file prevail on --cupn=CUPN option

2013-10-22 version 2.7¶

New features:

Monte-Carlo simulations are done in parallel on Windows platform too (needs R package snow)
if the option --seed=SEED is used, Monte-Carlo simulations are now reproducible even if run in parallel on multiple cores
for rank deficient Jacobian, the inequalities are now better enforced
starting value for maxstep parameter is set to \(10||p||\) instead of \(||p||\) where p is a vector of starting values for free parameters to fit.

Bug fixes:

fixed a bug preventing to report partial Monte-Carlo results if some simulations failed and some not
fixed a bug making to use all available cores instead of only one when NP was set to 1
fixed a fatal error when inequality enforcement fails
error and log messages during zero cross passes are made more explicit
fixed sending some error messages on standard output instead of .err file
when cumomer matrix is singular, fixed an error message about zero fluxes

2013-10-02 version 2.6¶

New features:

added option --sln (solution least norm) which applies ‘least norm’ to the whole solution vector of free parameters, not just to the increment vector (like --ln does)
a parallel calculation of multiple FTBLs is moved from python to R code. In such a way, some economies of repeated R starting up and library loading are made
when zero crossing is used (--zc=ZC) a third pass is added without any zc constraint.
added an option maxstep to control list of nlsic(). In some situations, it can make the convergence more stable at early iterations.

Bug fixes:

fixed a fatal error preventing from using BFGS optimization method
fixed an error in calculating reduced size of cumomer or EMU systems. It did not impact the results (at least for well defined network) but made calculations a little bit longer (reported by Stephane Mottelet, University of Compiegne, France).
a more explicit error message is generated when a given choice of free fluxes leads to a square but singular flux (stoechiometric) matrix.
some error messages were printed on standard output instead of .err file.

2013-06-28 version 2.5¶

New features:

an argument of the option --np=NP (number of processes) can be fractional, between 0 and 1 in which case the number of requested cores is calculated as NP*number_of_available_cores
in documentation, added a section describing some problematic cases and measures which could be undertaken to solve or to work around them. Few more field names in the output file are described (based on discussions with Yanfen Fu, University of Washington, USA)
missing values in measurements (NA as Non Available) are allowed in FTBL files.

Bug fixes:

fixed a fatal error if the rights of generated R file cannot be changed
fixed a bug for --ln (least norm) option when without inequalities, increments were not of least norm (reported by Stephane Massou, INSA of Toulouse, France)
fixed an algorithm used in --ln. Now for all inequality systems, both least residual norm and least solution norm are achieved (before, for some systems it was not the case). Due to this fix, we highly recommend to update to this version if you use --ln option
fixed a bug in “zero crossing” inequalities. Now, inequalities involving only constrained fluxes are canceled.

2013-04-11 version 2.4¶

New features:

number of parallel processes (in case of multiple FTBL files) is limited to a number of cores or to an argument of the --np option
some consistency controls were added on flux names in various FTBL sections.

Bug fixes:

fixed a bug in formatting some error messages during FTBL parsing;
fixed an accidental removing of kvh.py file from the previous release;
fixed non signaling to check .err file while some parsing errors did produce;

2013-03-28 version 2.3¶

New features:

external multicore R package is replaced by native parallel package;
convergence information of Monte-Carlo simulations is reported in the result file;
relative SD (rsd) in Monte-Carlo statistics is calculated as SD/abs(mean) and no more as SD/abs(estimated parameter);
if the number of really calculated samples in Monte-Carlo is less than 2, statistics are not calculated;
R code is self sufficient to be executed via source() function, even in parallel way;
with a new option --nocalc, R code is generated but not executed.

Bug fixes:

fixed concurrent access to a global variable in Monte-Carlo parallel execution;
fixed scope issue in Monte-Carlo simulations preventing from update of the current solution;
fixed some redundant warning messages;
fixed placement of .err and .log files if FTBL(s) is (are) given with subdirectories in their names.

2013-03-15 version 2.2.1¶

Bug fixes:

fixed a fatal error in Jacobian matrix construction when no measured fluxes are provided in FTBL file (reported by Yanfen Fu, University of Washington, USA);
in the User’s manual, added a naming convention for variable growth fluxes.

2013-03-13 version 2.2¶

New features:

if more than one FTBL file is given in argument to influx_s, all files are proceeded simultaneously in independent processes;
outliers in measurements can be automatically detected and excluded from parameter fitting.

Bug fixes:

fixed an error preventing Monte-Carlo results to be written if multicore package is not installed;
fixed a documentation error about \(\ln(M)\) in mynetwork.pres.csv file;
fixed warning resuming if there are many of them;
fixed some error message generation on FTBL parsing.

2013-02-15 version 2.1¶

New features:

in nlsic() a new field ‘retres’ is added to the list of returned values. It transfers “as is” the list returned by a last call to residual calculation function;
added a writing of generalized inverse of Jacobian to the result file;

Bug fix:

fixed a typo preventing Monte-Carlo statistics on forward-reverse fluxes to be written in the result file.

2013-02-05 version 2.0¶

New features:

metabolite pooling is modeled. Such pooling can appear due to compartmentation phenomenon or due to isomer coelution in chromatography. Starting from this version, metabolite concentrations can be part of fitted parameters;
adaptive backtracking algorithm is introduced to NLSIC algorithm;
history of convergence during minimization can be retrieved;
symbolic equations for dependent fluxes expressed as functions of free and constrained fluxes are generated by ftbl2cumoAb.py script;
METAB_MEASUREMENTS section is added to FTBL format;
added \(\chi^2\) test for evaluating the goodness of fit;
removed metab_scale field from OPTIONS section in FTBL format;
“dead end” internal metabolites are allowed in a network without being an output metabolite. As consequence, input-output fluxes must be explicitly declared as non reversible in the FTBL;
added optional EMU framework (--emu);
added optional series of starting points, fixed or randomly generated (--fseries, --iseries);
matrix construction is reworked and fortran code is removed. Now, no more Rtool installation is required for running influx_s;
some error messages are made more explicit and more precise;
outdated R package fUtilities is no more required;

Bug fixes:

fixed stoechiometric matrix construction when for a given metabolite; all fluxes are free or constrained;
fixed candidate propositions for free fluxes;
fixed standard deviation value for a DD/T field in PEAK_MEASUREMENTS section.

2011-10-11 version 1.0¶

Initial release. Main features:

NLSIC algorithm;
FTBL input format from 13CFlux project;
reduced cumomer set for cumomer balance equations;
sparse matrices;
usage of multicore R package for Monte-Carlo simulations on Unix platform;
usable on platforms having Python+numpy and R+some modules;
command line interface;
brief user’s and programmer’s documentation;
OpenSource (ECL) license.