CHANGES IN CHNOSZ 1.2.0 (2019-02-09)
------------------------------------

BUG FIXES

- Fix bug in nonideal() where "Zn" in formula was identified as charge.
  Thanks to Feng Lai for the bug report.

- Fix a bug where subcrt()$reaction$coeffs was incorrect for reactions
  involving minerals with phase transitions. Also ensure that the output
  reaction stoichiometry is correct for duplicated species in reactions.
  Thanks to Grayson Boyer for the bug report.

- For species in the revised HKF model, subcrt() now sets properties to
  NA where the density of H2O is less than 0.35 g/cm3, avoiding the
  output of bogus values in this region. Thanks to Evgeniy Bastrakov.

- For systems where no basis species is present in all formation
  reactions, and the user hasn't provided balance coefficients, stop
  with an error instead of setting the balance cofficients to 1.
  Thanks to Shuang Kong for an example calculation and Tucker Ely for
  a previous suggestion to produce an error here. The affected code is
  in balance(), an unexported function used in equilibrate() and
  diagram() (and now also solubility()).

NEW FEATURE: SOLUBILITY CALCULATIONS

- Add solubility(). Run this after affinity() to calculate the
  solubility of a solid or gas defined as the conserved basis species,
  which is involved in the formation of one or more dissolved species.

- Features include automatic detection of dissociation reactions and
  finding the final ionic strength for dissolution of a mineral into
  pure water (find.IS argument).

- find.IS depends on the new argument recall feature of affinity() (or
  mosaic() if that is used instead). This allows a calculation to be
  re-run with the same settings except for particular additions or
  modifications, in this case modified values of ionic strength.

- Revise demo/solubility.R to show solubility calculations for CO2(gas)
  and calcite as a function of T and pH.

NEW FEATURE: EXPANDED ACTIVITY COEFFICIENT CALCULATIONS

- Reorganize and expand options for activity coefficient calculations
  (set in thermo$opt$nonideal: Bdot, Bdot0, bgamma, bgamma0, or Alberty).
  The previous default, which corresponds to 'bgamma' (T- and
  P-dependent extended term parameter with single ion-size parameter),
  has been replaced by 'Bdot' (T-dependent extended term parameter and
  species-dependent ion-size parameter; see below).

- nonideal() with the 'Bdot' or 'Bdot0' equation uses specific
  ion-size parameters for different ions, in accord with the HCh package
  (Shvarov and Bastrakov, 1999). Parameters are from Table 2.7 of
  Garrels and Christ, 1965.

- nonideal() now calulates activity coefficients of neutral species,
  using the Setchénow equation. Whether the extended-term parameter in
  this equation is taken to be zero or is taken from the value for
  charged species (see above) is controlled by setting
  'thermo$opt$Setchenow' to bgamma0 (default) or bgamma.

- Add NaCl(), implementing a first-order calculation of the speciation
  of NaCl in water, taking account of activity coefficients and the
  reaction Na+ + Cl- = NaCl(aq).

OTHER NEW FEATURES

- Add dumpdata() for returning/writing all packaged thermodynamic data
  (including default database and optional data files). The file is
  also available on the website (chnosz.net/download/alldata.csv).

- mosaic() has been rewritten to handle more than two changing groups
  of basis species.

- mosaic() gets a new argument 'mixing' (default TRUE), indicating
  whether the Gibbs energy of ideal mixing should be included in the
  calculations with blend = TRUE. As a test of this change, an example
  using mosaic() to calculate the pH-dependent solubility of calcite
  has been added to solubility.Rd.

DOCUMENTATION

- Add demo/gold.R for calculations of Au solubility in hydrothermal
  chloride and sulfide solutions (based on diagrams from Akinfiev and
  Zotov, 2001, Stefánsson and Seward, 2004, and Williams-Jones et al.,
  2009). This depends on the revised nonideal() and new NaCl() functions
  described above.

- anintro.Rmd: add section on solubility calculations. Add cuprite to
  mosaic diagram example, and note about implications of changing
  balance coefficients to 1. Add 'mklinks.sh' script for adding links
  to CHNOSZ and R documentation.

- Add demo/bison.R (average oxidation state of carbon of metagenome-
  derived proteins in different microbial phyla at Bison Pool)

- demo/saturation.R replaces the previous activity_ratios.R, and
  includes not only activity ratios on the axes but also saturation
  limits for calcite, dolomite, magnesite, and brucite in the
  H2O-CO2-CaO-MgO-SiO2 system.

- Add demo/glycinate.R showing logK of complexation of glycinate with
  divalent and monovalent metals.

- Remove vignette wjd.Rnw (Winding journey down in Gibbs energy).

THERMODYNAMIC DATA

- The Berman data (Berman, 1988 and later additions) have replaced the
  SUPCRT92 data (based on Helgeson et al., 1978) for most minerals in
  the default database (i.e. the one loaded by data(thermo)). Only
  native elements, sulfides, halides, sulfates, and selected carbonates
  and oxides from SUPCRT92 that do not duplicate any in the Berman
  dataset are loaded by default.

- The superseded SUPCRT92 minerals (primarily the silicates and
  Al-bearing minerals, as well as calcite, dolomite, hematite, and
  magnetite) are in the file extdata/OBIGT/SUPCRT92.csv, which can be
  loaded with add.obigt("SUPCRT92"). This is used in some examples for
  comparing the datasets (see ?berman and demo("go-IU")), but should not
  be needed by most users.

- In thermo$obigt, the minerals using the Berman equations are
  identified as those having NA values for all thermodynamic properties
  (columns 8-20). These minerals have the state of "cr" (not "cr_Berman"
  as before).

- The heat capacity equation in berman() has been expanded to
  Cp = k0 + k1T-1/2 + k2T-2 + k3T-3 + k4T-1 + k5T + k6T2 (as used in
  winTWQ).

- Multipliers on values in the Berman data files (extdata/Berman/*.csv)
  have been removed, except for 10^5, 10^5, 10^5, and 10^8 on v1 to v4,
  to be consistent with the winTWQ data format.

- berman() now reads data from a user-supplied file specified in
  thermo$opt$Berman.

- Add mineral data from Berman and Aranovich, 1996 to
  extdata/Berman/testing and an example of loading these data via
  add.obigt() and thermo$opt$Berman (see ?berman).

- Correct swapped v1 <-> v3 and v2 <-> v4 in extdata/Berman/DS10.csv.

- To reduce confusion that "d5" might be a coefficient in the same
  equation as d0 .. d4, rename "d5" (constant that scales disordering
  enthalpy to volume) to "Vad" (notation used in Theriak-Domino manual).

- Change abbreviation of grossular to Grs.

- Make names for dicarboxylic acid anions from Shock, 1995 (in
  organic_aq.csv) consistent with those used in later papers, e.g.
  H-succinate and succinate-2 for the -1 and -2 charged ions. Remove
  duplicated entries for H-succinate and succinate-2 from biotic_aq.csv.

- Update G, H, and S of celestite using properties of the dissolution
  reaction given by Reardon and Armstrong, 1987.

- Add scheelite (CaWO4) and ferberite (FeWO4) using data adopted by
  Wood and Samson, 2000.

- Update aqueous Au species with data from Akinfiev and Zotov, 2001
  and 2010, and Pokrovski et al., 2014.

- Move SUPCRTBL updates (As and Al minerals and aqueous species and
  SiO2(aq)) into default database.

- Values for SiO2(aq) are taken from Apps and Spycher, 2004. Thanks
  to John Apps for providing the document.

- Recalculate GHS for HSiO3- by adding difference from SiO2(aq)
  (Sverjensky et al., 1997) to updated values for SiO2(aq).

- Add test-recalculate.R to check that recalculated values are correctly
  entered in OBIGT.

- Move superseded data for aqueous Au, As, and Al species, SiO2 and
  HSiO3- to OBIGT/SLOP98.csv.

- Add or update glycine, diglycine, and triglycine (zwitterions and
  ions), and diketopiperazine, [Gly] and [UPBB] groups from Kitadai,
  2014.

- Add data for metal-glycinate complexes from Azadi et al., 2019,
  superseding Shock and Koretsky, 1995.

- For reproducing previous calculations, superseded data for [Gly] and
  [UPBB], as well as [Met] (earlier superseded by LaRowe and Dick,
  2012), and metal-glycinate complexes have been moved to
  OBIGT/OldAA.csv.

DIAGRAMS

- Lines in 1-D diagram()s can optionally be drawn as splines using the
  method for splinefun() given in the 'spline.method' argument (the
  default of NULL means no splines).

- Add 'srt' argument to diagram() (rotation of line labels).

- Export thermo.axis(), as it is useful for adding major and minor tick
  marks after (above) other plot elements such as legends.

- diagram(): rename 'what' argument to 'type'.

- diagram(): add new type of diagram, 'saturation', which is used to
  plot saturation lines for minerals (where their affinity equals
  zero).

- thermo.plot.new() and, by extension, diagram(), now "listen" to the
  setting of par("mar"). However, if a new plot is being made, or the
  device has the default setting of mar from par(), mar is (re)set to
  the default value in thermo.plot.new().

- diagram(): numeric 'names' now indicates subset of species names to
  include as labels on the plot.

- diagram(): new argument 'family' to select the font family for species
  names (labels on the plot). New arguments 'bold' and 'italic' apply to
  formatted names (e.g. chemical formulas), and 'font' applies to those
  without formatting.

- 'lines' is added to the return list of diagram(), giving the
  coordinates of lines (field boundaries) on 2-D diagrams (these are
  taken from the output of contourLines()).

COMPUTATIONAL OPTIONS

- Add 'exceed.rhomin' argument to subcrt() and affinity() to enable
  output of properties for species in the revised HKF model below 0.35
  g/cm3.

- In equilibrate(), accept a length > 1 'normalize' argument to
  normalize the chemical formulas of only the selected species.

- Add thermo$opt$maxcores (default 2) to specify maximum number of
  cores for parallel calculations with palply().

- Keywords in basis(): Change 'CHNOPS+' to use O2 instead of e-, and add
  'CHNOPSe' and 'MgCHNOPSe' for sets of basis species that have e-.

- Add 'keep.duplicates' argument to thermo.refs(). Set it to TRUE to
  output a single primary reference for each species, keeping any
  duplicated references (but not including any secondary references in
  thermo$obigt$ref2). Thanks to Evgeniy Bastrakov for the suggestion.

USABILITY ENHANCEMENTS

- To provide better diagnostics for other apps using CHNOSZ, warning
  messages produced by subcrt() are now available in the output of
  affinity(), under 'sout$warnings'.

- Change internal variable names in subcrt() for better readability
  (sinfo -> ispecies, inpho -> iphases, sinph -> phasespecies).

- info(): display mineral names instead of just "cr" in messages (e.g.
  for info("SiO2")).

- info(): species with no matches in default database are also
  searched in optional data files.

- Add C implementation of counting occurrences of all letters in a
  string (src/count_letters.c) to speed up operation of count.aa().

- read.fasta(): add support for file connections created using
  archive::archive_read (https://github.com/jimhester/archive).

- The arguments in expr.species() have been reorganized for more
  flexible and concise usage.

- Reduce message clutter for diagrams showing water stability limits.

- In subcrt(), improve error message for invalid property names.

- In affinity(), make invalid variable names an error.

- Add 'return.sout' argument to affinity(), to return just the values
  calculated with subcrt().

CHANGES IN CHNOSZ 1.1.3 (2017-11-13)
------------------------------------

MAJOR CHANGES:

- Add berman() function and extdata/Berman/*.csv files for calculating
  thermodynamic properties of minerals using equations of Berman, 1988.

- Calculations related to Berman's (1988) Figs. 1 and 2 for the lambda
  transition of quartz are available in the new demo lambda.R.

- Add functions implementing the Deep Earth Water (DEW) model
  (Sverjensky et al., 2014): water.DEW() and its supporting functions
  calculateDensity(), calculateGibbsofWater(), calculateEpsilon(),
  calculateQ().

- The computational setting for water (thermo$opt$water) can now be set
  using water("DEW"), water("IAPWS"), etc.

- Usage of the DEW model is shown in the new demo DEW.R. This demo also
  depends on the Berman equations (above) and, for the last diagram in
  the demo, the following *four* NEWS items:

- In equilibrate(), it is now possible to combine affinity calculations
  with variable activity of the balancing basis species (loga.balance).
  For example, in the last plot of the DEW demo, the calculation
  involves simultaneously varying temperature and logfO2 (given as
  arguments to affinity()) as well as total concentration of carbon
  (given by the loga.balance argument in equilibrate()).

COMPUTATIONAL IMPROVEMENTS:

- The 'alpha' argument of diagram() can be set to 'balance' to scale
  the values by the balancing component. This is useful for making
  "percent carbon" plots for systems where the species have different
  carbon numbers.

- nonideal() now has three methods for calculating activity
  coefficients: 'Helgeson' and 'Helgeson0', which utilize the
  Debye-Huckel equation with parameters as described by Helgeson, 1969,
  and 'Alberty' (the only method previously available). The 'Helgeson'
  method depends on the following *two* NEWS items (but the 'Helgeson0'
  method omits the B-dot term):

- All three water options (SUPCRT92, IAPWS95, DEW) can be used to
  calculate 'A_DH' and 'B_DH', i.e. A and B coefficients in the extended
  Debye-Huckel equation as described by Helgeson, 1969.

- Add Bdot() to calculate the "B-dot" (or b_gamma) extended term
  parameter for activity coefficients in NaCl solutions at high
  temperature and pressure (Helgeson et al., 1981), including
  very high pressures (> 10 kbar) based on extrapolations from Manning
  et al., 2013.

- For minerals with phase transitions (states 'cr2' 'cr3' etc.) in
  thermo$obigt (i.e. the Helgeson minerals), it is now possible to use
  the minerals in basis(), species(), affinity() with proper accounting
  for transition effects. Just set the state to 'cr' or omit it,
  e.g. species("quartz").

- Implement SUPCRT92's handling of variable volume for quartz and
  coesite. Calculations for other minerals still assume constant
  volume of each phase.

- Calculations of the g function are now enabled for DEW (with
  pressure derivative) and IAPWS-95 (no derivatives included).

- water.lines() now works for diagrams of Eh, pe, logfO2, logaO2,
  logfH2, or logaH2 vs pH, T, or P. It is possible to have T or P on
  either the x- or y-axis.

- Variables that are in the aguments to affinity() now override any
  buffers previously specified using basis().

DATABASE UPDATES:

- Add data(OBIGT) command to reset only the thermodynamic database
  (OBIGT) to its default. data(thermo) is still used to reset
  *everything* (the database and all computational and system settings).

- In add.obigt(), using 'file' without a file suffix now can be used to
  locate corresponding files (such as DEW_aq.csv or SUPCRTBL.csv) in the
  extdata/OBIGT package directory.

- add.obigt() gets new argument 'species' for selecting species to add.

- Add Berman_cr.csv with names and formulas for minerals whose
  thermodynamic properties are calculated using the Berman, 1988
  equations. The state "cr_Berman" is used to distinguish these
  minerals from the Helgeson data. The parameters themselves are
  stored in extdata/Berman/*.csv.

- Add DEW_aq.csv with aqueous species data from the DEW spreadsheet
  (May 2017 version). Species with data that do not differ from the
  default database in CHNOSZ are not included. Currently, detailed
  reference keys and descriptions (in thermo$refs) are given for
  selected species only. [Optional Data; not loaded by data(thermo)]

- Add SUPCRTBL.csv with mineral and aqueous species data compiled by
  Zimmer et al., 2016. Data were taken from original references or
  from spronsbl.dat (http://www.indiana.edu/~hydrogeo/supcrtbl.html).
  The new demo/go-IU.R uses these data. [Optional Data]

- Update HKF parameters of adenine based on experimental heat capacity
  and volume data (Lowe et al., 2017). Add demo/adenine.R comparing
  the updates with the previous data (LaRowe and Helgeson, 2006).

- Remove H2O(gas) from the database (taken from SUPCRT92), as it gives
  inconsistent results for metastable steam (Zavarin et al., 2016,
  LLNL-TR-701407, doi: 10.2172/1325873).

- Change the formula of the electron from "Z0-1" (parsed as "zero Z
  with minus one charge") to the equivalent but nicer-to-read "(Z-1)"
  (parsed as "minus one Z"; the parentheses cause "-1" to be interpreted
  as the coefficient on Z instead of the charge in the formula).

- Add data for hedenbergite, andradite, ferro-actinolite, grunerite, and
  ilvaite from Delgado Martin and Soler i Gil, 2010 (using the Berman
  equations).

OTHER CHANGES:

- Add test-logmolality.R to demonstrate transformation of variables from
  activity to molality in the main workflow; refer to this test and
  describe these transformations in anintro.Rmd; remove "experimental"
  labeling of activity coefficient calculations in CHNOSZ-package.Rd.

- Add 'use.molality' argument to axis.label() and supporting functions.
  So that the labels on diagrams are changed from activity (a) to
  molality (m), this argument is set to TRUE by diagram() when IS is
  supplied as an argument to affinity().

- Add demo TCA.R for standard Gibbs energies of steps of the
  citric acid cycle (Canovas and Shock, 2016).

- Components of subcrt() output indicating the stable polymorph of
  minerals with phase transitions are now named 'polymorph', not
  'state'.

- Add maxdiff() and expect_maxdiff() for calculating and testing the
  maximum absolute pairwise difference between two objects.
  expect_maxdiff() is used in tests/testthat/*.R as a more
  straightforward alternative for checking differences than using
  'tolerance' in expect_equal().

- Add 'tol' argument to equil.reaction() (convergence tolerance for
  uniroot()).

- Add thermo$opt$Berman to signal info(), subcrt(), etc. to prefer
  minerals in Berman dataset over Helgeson dataset (default: FALSE).

REFACTORING AND CLEANUP:

- In hkf() and cgl(), combine 'ghs' and 'eos' arguments into single
  argument named 'parameters'.

- Remove 'H2O.PrTr' and 'domega' arguments from hkf(). Properties of
  water at Pr and Tr and the logic for when to find derivatives of omega
  are now calculated within the function, thereby simplifying the
  function call from subcrt().

- Add 'H2O.props' argument to hkf(). This lists the properties of water
  to calculate (at P, T); these are now included in the results, in
  order to avoid duplicating the calculations in subcrt().

- In water.* functions, rename the "diel" variable to "epsilon".

- To save space, taxid_names.csv has been trimmed to hold only those
  taxids listed in extdata/bison/gi.taxid.txt.

- Remove Eco.csv.xz and rename more.aa() to yeast.aa().

- Remove transfer() and associated functions draw.transfer(), apc(), and
  feldspar().

- Remove read.expr() and extdata/abundance/ISR+08.csv (protein abundance
  in E. coli).

- Reduce the size of obigt.html by ~450 KB. This was done by 1) using
  rmarkdown with files in rmd/h/bootstrap-3.3.5/fonts/ replaced with
  empty files (291 KB), and 2) changing the theme from "cosmo" to
  "default" (151 KB).

- Remove extdata/OBIGT/CHNOSZ_aq.csv and move the entry for
  pseudo-H4SiO4 to inorganic_aq.csv.

- Reorganize concept index entries in Rd files: Main workflow, Extended
  workflow, Thermodynamic data, Thermodynamic calculations, Water
  properties, Protein properties, Other tools, Utility functions.

- Remove anim.TCA() and anim.carboxylase(); move the latter code to
  demo/carboxylase.R.

CHANGES IN CHNOSZ 1.1.0 (2017-05-04)
------------------------------------

DOCUMENTATION:

- Replace anintro.Rnw (Sweave) with rewritten anintro.Rmd (using knitr
  with Tufte style).

- Add obigt.Rmd (Thermodynamic data in CHNOSZ).

- Add eos-regress.Rmd (Regressing thermodynamic data), including
  section on Making a pseudospecies: H4SiO4.

- New demos: bugstab.R (potential diagrams for microbial proteins in
  colorectal cancer), activity_ratios.R (mineral stability diagrams
  with activity ratios on the axes), Shh.R (affinities of
  transcription factors relative to Sonic hedgehog).

- Converted demos: move examples from help pages to protein.equil.R,
  add.obigt.R, and affinity.R.

- Updated and removed demos: update buffer.R (logfH2 set by mineral
  buffers and aqueous species) and rename old buffer.R to protbuff.R
  (proteins as buffers); remove nucleobase.R.

NAMESPACE:

- The NAMESPACE now exports selected functions, instead of all of them.

- The following functions are no longer exported, nor documented in the
  Rd files: thermo.axis, optimal.index, extremes, caller.name, lsub,
  lsum, pprod, energy, energy.args, A.ionization, can.be.numeric, s2c,
  c2s, eos.args, TP.args, state.args, info.approx, info.character,
  info.text, info.numeric, count.charge, count.formulas, buffer,
  mod.basis, put.basis, preset.basis, preset.logact, species.basis,
  envert, outvert, which.balance, gfun, IAPWS95.residual,
  IAPWS95.idealgas, obigt2eos, group.formulas, get.formula, get.objfun,
  balance.

CHANGES TO diagram():

- For the `groups` argument, activities are multiplied by the balance
  coefficients before the summation.

- Draw smoother boundary lines (with contour(); `lty` is usable), at the
  expense of not being precisely aligned with the grid used for plotting
  colors. For the old behavior, use `dotted=0`.

- Default of `legend.x` changed to NA (instead of making a legend, put
  labels near the lines).

- Field labels: apply formatting (subscripts and superscripts) to names
  that parse as chemical formulas; use `format.names=FALSE` to disable.

- Chemical formulas that are part of the axis labels (log activity or
  fugacity) now have formatting applied.

- Add arguments `adj` and `dy` for x-alignment and y-offset of line
  labels.

- Add arguments `fill.NA` (color of empty areas) and `limit.water`
  (assign NA to areas beyond water stability limits on Eh-pH and pe-pH
  diagrams).

NEW FEATURES:

- Add ZC.col() for generating a red-grey-blue color scale from
  numeric values.

- Add "QEC" as a keyword for preset species in basis() (glutamine,
  glutamic acid, cysteine, H2O, O2).

- More flexible parsing of chemical formulas for ZC() and other
  functions; e.g. `ZC(colMeans(protein.formula(1:4)))` now works.

- read.fasta() extracts information from UniProt FASTA headers
  (accession, name, organism) into columns of the output data frame
  (protein, abbrv, organism).

- Add `...` argument to thermo.plot.new() (additional arguments for
  par()).

- With new default arguments, thermo.axis() only plots the axis tick
  marks and lines; this is used in diagram() to redraw the axes on
  filled diagrams.

- seq2aa() removes newlines and whitespace before counting the
  letters in the sequence.

- read.fasta(): change argument `i` to `iseq`; this is used to select
  particular sequences to read from the file.

- Add syslab() and ratlab() to make formatted expressions for
  sets of thermodynamic components and activity ratios.

- Add GHS_Tr() to calculate G, H, and S at Tr (298.15 K) of phases
  stable at high temperature, using provided enthalpies of transition
  and database values of transition temperatures and heat capacity
  parameters, and GHS at Tr of the low-temperature phase.

- Add extdata/supcrt/ directory with files read.supcrt.R, compare.R,
  newnames.csv; used to compare data in SUPCRT and slop files
  (http://geopig.asu.edu/?q=tools) to data in thermo$obigt.

NEW AND OLD DATA:

- Split OBIGT.csv and OBIGT-2.csv into multiple files according to type
  of substance (inorganic, organic, or biotic). All files are now under
  extdata/OBIGT/. Conflicting data from OBIGT-2.csv were removed (see
  below).

- Citations in OBIGT now have a suffix (e.g. [S92] or [S98]) if the data
  were first present in sprons92.dat from SUPCRT, slop98.dat, or later
  versions of the slop data files (http://geopig.asu.edu/?q=tools).

- biotic_aq.csv: Add data for citric acid cycle metabolites from
  Canovas and Shock, 2016.

- CHNOSZ_aq.csv: Add H4SiO4 pseudospecies (pseudo-H4SiO4); see
  calculations in eos-regress.Rmd.

- organic_aq.csv: Add data for aqueous phenanthrene and methylphenanthrene
  from Dick et al., 2013.

- Remove superseded data previously in OBIGT-2.csv: citric acid cycle
  metabolites from Dalla-Betta and Schulte, 2009.

- Remove superseded data previously in OBIGT-2.csv: methionine and [Met]
  from Dick et al., 2006. Examples that require the old group additivity
  parameters can modify the database using:
  `mod.obigt("[Met]", G=-35245, H=-59310)`.

- Move two datasets that conflict with the primary database, but are
  used in tests, examples, or vignettes, from OBIGT-2.csv to
  extdata/thermo (Ste01.csv and BZA10.csv).

- Remove other conflicting datasets (i.e. that provide different data
  for a species already in the database) previously listed in
  OBIGT-2.csv (21 sources).

- Replace GHS for iron (c2, c3) calculated in CHNOSZ with values derived
  from sprons92.dat.

- Replace data for sulfur (cr) from Gurrieri, 1996, personal
  communication with that "as used by" McCollom and Shock, 1997 (from
  slop98.dat).

- Update selected nonelectrolytes with data from Plyasunov and Shock,
  2001, as listed in slop07.dat (http://geopig.asu.edu/?q=tools).

- extdata: Add fasta/rubisco.fasta and cpetc/*.csv files SS98_Fig5a,
  SS98_Fig5b, PM90 (all used in anintro.Rmd).

BUG FIXES:

- Correct charge (-2) of NAD(red)-2 in biotic_aq.csv. Thanks to Peter
  Canovas.

- Correct charge (-2) of MgATP-2.

- subcrt() returns `loggam` using the common logarithm; add
  test-nonideal.R to check for consistency between loggam and logK
  values returned by subcrt(). This also fixes the issue previously
  noted for the first example in ?nonideal. Thanks to David T. Wang for
  the bug report and test.

- NaN values from equilibrate() are now preserved by diagram(),
  producing empty (NA) fields rather than being mistakenly labeled with
  the first species. Thanks to Grayson Boyer for the bug report.

OTHER CHANGES:

- Remove msgout(), and replace previous calls to that function with
  message() from base R. As a result, the messages don't appear in
  Sweave vignettes, but can be turned on or off in knitr vignettes.

- Remove HTCC1062.faa.xz.

- Remove the bold formatting for "A" in expr.property(); "A" is
  italicized like most other uppercase letters.

- Rename browse.refs() to thermo.refs(); remove URL browsing (except for
  summary table).

- New function pinfo() merges functionality of old iprotein() and
  ip2aa(). Add `regexp` argument to control whether matches are made
  using a regular expression.

- Rename aa2eos() to protein.obigt().

- Remove anim.plasma() and AA03.csv.

- Remove stress() and stress.csv; move data from Tai et al., 2005 (used
  in an example in ?read.expr) to TBD+05.csv.

- Remove read.aa() - replaced by read.csv() with as.is=TRUE.

- Remove grep.file().

- Remove sideeffects.Rd (functions with side effects are now identified
  with red color in anintro.Rmd).

CHANGES IN CHNOSZ 1.0.8 (2016-05-28)
------------------------------------

- Add gypsum to OBIGT.csv.

- Add MgSO4 and hydrated Mg sulfate minerals to OBIGT-2.csv (kieserite,
  starkeyite, hexahydrite and epsomite from Grevel and Majzlan, 2009).

- Add demo dehydration.R. This demo requires the RSVGTipsDevice package,
  which is currently unavailable on Windows.

- Add ... argument to EOSvar() and EOSregress(). Allows specifying new
  variables that are written as a function of arbitrary properties
  (regression variables still must be a function of at least T and P).

- EOSvar() now returns the negative of the 'QBorn' variable so that
  values of omega obtained from regressions of volume data have the
  correct sign in the HKF equations.

- Use given/family naming scheme in inst/CITATION.

- makeup() now works even if 'thermo' object is not available (useful
  for other packages importing this function). Thanks to Nir Shahaf for
  the suggestion.

- So that byte-compiled package code works, subcrt() no longer uses
  caller.name() for flow control. In its place, a new argument
  'check.Ttr' has been added, used to bypass the checks of transition
  temperature in calls from dPdTtr(). Thanks to Tomas Kalibera for the
  suggestion.

CHANGES IN CHNOSZ 1.0.7 (2015-11-19)
------------------------------------

- Add functions usrfig() (get figure limits in user coordinates) and
  label.figure() (add label to figure outside of plot region).

- Add demo copper.R (complexation of copper with glycine, uses
  mosaic()).

- Using new supporting function ibasis(), swap.basis() and mosaic()
  now allow basis species to be specified by name, not only formula.

- diagram() now has '...' argument to pass additional options to
  plot() (useful with diagram(tplot=FALSE, ...)).

- Update vignette 'equilibrium.Rnw' with better definitions of concepts,
  organization of functions, and examples and applications. Now uses
  knitr.

- Remove demo diagram.R (concepts better shown in equilibrium.Rnw).

- Rename basis.matrix() to basis.elements().

- Argument of species.basis() defaults to current species indices.
  ( species.basis() %*% basis.elements() returns elemental composition
  of currently defined species.)

- Move demos phosphate.R and nonideal.R to help page nonideal.Rd.

- Add options to thermo$opt: 'ideal.H' and 'ideal.e'. Default TRUE tells
  nonideal() to set activity coefficients of proton and electron to
  zero.

- Add 'method' argument to equilibrate() (can be 'reaction' or
  'boltzmann').

- For compatibility with a change in R-devel ("aperm(a, *) now preserves
  names(dim(a))"), modify slice.affinity() to preserve names of
  dimensions, and use check.attributes=FALSE or check.names=FALSE in
  some tests.

CHANGES IN CHNOSZ 1.0.6 (2015-10-19)
------------------------------------

- Rewrite rho.IAPWS95() to be able to invert density from IAPWS-95
  equations for a more extensive range of T,P values.

- Update R dependency to R-3.1.0, needed for 'extendInt' argument
  of uniroot() (used by rho.IAPWS95()).

- Add demo/density.R to show density of H2O calculated using
  rho.IAPWS95() (optionally using water.SUPCRT92()).

- Remove the relatively trivial demo/cordierite.R.

- Update wrapper function for parallel calculations palply() (export of
  variables, used in demo/density.R) and add tests in
  test-util.program.R.

- Tests that initiate calls to palply() (and therefore makeCluster())
  failed with 'cannot open file 'startup.Rs': No such file or directory'
  Fixed by adding Sys.setenv("R_TESTS" = "") to test-all.R.
  Issue discussed here:
  https://github.com/hadley/testthat/issues/129
  https://github.com/hadley/testthat/issues/144

- Fix bugs in which.pmax() that prevented proper assignment of
  attributes in output, and functionality for lists of length 1.

- mosaic() now multiplies affinities by _relative_ abundances of basis
  species when blend=TRUE. Also, fix bug where using blend=TRUE gave
  wrong results due to incorrect calculation of total activity of basis
  species. Thanks to Grayson Boyer for the bug reports.

- Add grDevices, graphics, stats, utils to Imports: in DESCRIPTION, and
  add specific importsFrom lines in NAMESPACE.

- Fix incorrect entry for entropy of aqueous methionine and [Met] in
  OBIGT.csv. Thanks to Apar Prasad for reporting this.

- Some fixes for compatibility with new version of testthat (increase
  tolerance for one test in test-eos.R because of corrected alignment
  actual/expected values in equals(); move inst/tests to
  inst/test/testthat; use test_check instead of test_package in
  test-all.R (for R CMD check)).

CHANGES IN CHNOSZ 1.0.5 (2015-05-19)
------------------------------------

- Following "Writing R Extensions", remove uncommented \donttest{} code
  from Rd files. Some examples have had \donttest{} markup removed
  without other modification, some non-representative examples have
  been deleted, and some longer examples have been moved to demo files.

- Some previous \donttest{} examples have been moved to demos 'findit.R'
  (replaces old demo), and to new demos 'revisit.R', 'ionize.R',
  'buffer.R', 'wjd.R', 'yeastgfp.R', 'diagram.R'.

- Rd files now use standard R \keyword{}s ('package' and 'utilities');
  non-standard keywords have been changed to \concept{}s (see CHECKLIST
  for description, and CHNOSZ-package.Rd for overview).

CHANGES IN CHNOSZ 1.0.4 (2015-05-18)
------------------------------------

NEW FEATURES:

- Add mosaic() function for affinity calculations with changing basis
  species. See also new 'mosaic.R' demo.

- Add thermo$opt$varP option for subcrt() to calculate Gibbs energies
  of gases using a variable-pressure standard state. Thanks to Kirt
  Robinson for the submitting the problem that led to this change.

- Add 'solubility.R' demo, and add calcite solubility plot to
  anintro.Rnw.

DATA CONSISTENCY:

- Move aqueous Ti(OH)4 and H4SiO4 to OBIGT-2.csv.

- Add warning about data consistency and use of examples to
  CHNOSZ-package.Rd and anintro.Rnw.

OTHER CHANGES:

- When making filled color diagrams, diagram() calls image() with
  useRaster=TRUE. This speeds up plotting, eliminates artifacts (i.e.
  blank lines) at some resolutions, and greatly reduces the size
  of PDF files.

- diagram() returns 'lx', 'ly' and 'is' for calculated positions of
  labels and indices of those species in predominance diagrams,
  including when the labels aren't plotted (by setting names=FALSE).

- In energy(), add missing 'IS' argument in call to subcrt(). Thanks to
  Grayson Boyer for the bug report.

- In nonideal(), skip proton and electron. Add test for calculations
  at constant and variable IS to test-util.affinity.R.

- Modify hotspring.Rnw to use knitr, and add example calculations of
  relative abundances of microbial phyla from Dick and Shock, 2013.

- Add files with average amino acid compositions of proteins from Bison
  Pool hot spring grouped according to annotation keyword (DS11.csv)
  (moved here from protein.csv) or phylum assignment (DS13.csv).

- water.lines() gets 'O2state' argument to specify state of O2.

- Add 'as.residue' argument to diagram(), to divide reactions by
  balance coefficients without rescaling to whole species formulas
  (cf. 'normalize' argument which uses rescaling).

- In equilibrate(), change name of 'stay.normal' argument to
  'as.residue'.

- In test-diagram.R, test that same diagrams are produced using
  'normalize' in equilibrate() or diagram().

- In test-diagram.R, test that same diagrams are produced using
  'as.residue' in equilibrate() or diagram().

- Add 'tplot' argument to diagram() (change to FALSE to use R's default
  plot set-up).

- R version dependency changed to (>= 3.0.0).

CHANGES IN CHNOSZ 1.0.3 (2014-01-12)
------------------------------------

- Updated extdata/protein/Sce.csv.xz using Saccharomyces Genome
  Database protein_properties.tab and SGD_features.tab dated
  2013-08-24.

- more.aa() includes SGDID and gene name in the abbrv and ref columns.

- In extdata/refseq, scripts and data files were updated for NCBI
  Reference Sequence (RefSeq) release 61 (2013-09-09). Code was
  adapted to deal with WP multispecies accessions.

- read.fasta() gets new argument 'id'; when supplied, it is used for
  the protein names in the output, in place of those read from the
  FASTA headers.

- When reading protein names from the FASTA headers, read.fasta()
  stops only at the first space, not space or underscore as before.

- info() no longer specially sets state of "O2" to gas. The name
  "oxygen", or the combination ("O2", "gas"), can be used to retrieve
  data for the gas.

- In thermo$obigt, names of gases (e.g. "oxygen") are used only for
  the gaseous species; names were removed from dissolved species in
  rows 63-79. Both gases and aqueous species can continue to be
  referenced by their chemical formulas.

- In read.expr(), allow multiple filter specifications.

- revisit() has new argument loga0, a single vector of base-10
  logarithms of activities of species used to calculate the base-2 log
  ratio ( log2(a1/a0) ).

- Updated tests to be compatible with testthat version 0.8 (in
  development). Some tests now check for e.g. an error *and* warning
  produced by a single function call.

CHANGES IN CHNOSZ 1.0.1 (2013-07-04)
------------------------------------

- Allow IAPWS-95 calculations to be used in subcrt(): in 
  water.IAPWS95(), rename internal function from 'epsilon' to 'diel';
  in water(), return upper-case versions of names of properties.

- rho.IAPWS95() has modified search interval to give results at
  T < 250 K and P < 300 bar. Thanks to Marc Neveu.

- In water.AW90(), analytical solution to polynomial equation is
  now used to calculate dielectric constant. Thanks to Marc Neveu.

- count.aa() and read.fasta() handle DNA sequences, with new argument
  'type' (protein or DNA).

- nucleicacids() split into nucleic.formula() and nucleic.complement().

CHANGES IN CHNOSZ 1.0.0 (2013-03-28)
------------------------------------

MAJOR USER-VISIBLE CHANGE:

- On attaching the package, the user is asked to load the 'thermo'
  object, containing thermodynamic data and system settings, using
  data(thermo). This is required because packages are not permitted to
  alter the search path (but the user may).

OTHER CHANGES:

- Fix calculation of free energy derivative in wjd().

- Add 'stay.normal' argument to equilibrate().

- If obigt$G is available, hkf() returns this value, not NA, at Tr, Pr.

- mod.obigt() defaults to taking chemical formula from the species name,
  and checks for validity of formula.

- Add example for LYSC_CHICK to protein.info.Rd.

- Enable DGtr in findit(), 1-D DGtr plot in revisit().

- Disable a check in valTP function of H2O92D.f to allow properties of
  H2O to be calculated below 0.01 degrees C, to -20 degrees C.

- Remove thermo$water and thermo$water2 storage of previous results;
  they gave no significant speed gain in running examples.

- Split IAPWS95.Rd out of water.Rd.

- expr.* functions now do not wrap their values in as.expression()
  (makes compounding expressions, e.g. with substitute(), easier).

- Rename dl.aa() to uniprot.aa().

- Add start and stop arguments to count.aa(), read.fasta(),
  uniprot.aa().

- add.protein() replaces amino acid compositions for existing
  proteins with the same name.

- Examples of calculation of affinity of formation of CSG_METVO
  (following Dick and Shock, 2011) added to protein.info.Rd.

- Use consistent names for water properties (Speed, diel, QBorn, ...).

- Add water.props() to get names of properties of water.

- Remove 'isat' argument from water.SUPCRT92(); function now accepts
  'Psat' as value for 'P' argument.

- Separate rho.IAPWS95() from water.IAPWS95().

- In addition to the original regression variables, EOSvar() recognizes
  names for available properties in water(), or can use the name to
  get a user-defined function of temperature and pressure.

- Simplify EOSplot() somewhat (don't group data by pressure ranges).

- EOSlab() gets label from attribute (if present) of user-defined
  function.

- Modify test-util.data.R and wjd.Rd to pass R CMD check using R
  configured with --disable-long-double .

- The name of the environment affected by data(thermo), and used in many
  functions, is changed from CHNOSZ:thermo to CHNOSZ.

- Remove read.supcrt() and write.supcrt().

- guess() now defaults to "stoich" method, not "central". (Needed since
  limSolve package is not found during R-Forge checks on Windows.)


CHANGES IN CHNOSZ 0.9-9 (2013-01-01)
------------------------------------

MAJOR CHANGES:

- Split the functionality of diagram() into equilibrate() and diagram().
  Old workflow: a <- affinity(); d <- diagram(a) .
  New workflow: a <- affinity(); e <- equilibrate(a); d <- diagram(e) .
  Old workflow is still usable for plotting the values of affinity, or
  for making predominance diagrams using the maximum affinity method.

- The 'thermo' object, which holds the thermodynamic database,
  and system definitions (made by the user), is now placed in an
  environment named 'CHNOSZ:thermo' on loading the package. Therefore, 
  instances of '<<-' in the code now refer to this environment
  instead of the global environment.

- Create a set of tests in 'inst/tests', particularly for functions that
  have been modified during this development cycle, and add a Suggests 
  dependency on 'testthat'.

- Move the code for the temperature and pressure derivatives of the "g"
  function (related to the solvation parameter omega) to a new function 
  gfun(); incorporate some fixes and a series of test_that() tests.
  There is some impact on the calculated Gibbs energies of charged
  species.

NEW FUNCTIONS:

- Add wjd() implementing the steepest descent algorithm for free energy
  minimization described by White et al., 1958. Also add supporting 
  functions element.potentials(), invertible.combs() for finding 
  linearly independent combinations of rows of a matrix, 
  is.near.equil(), and run.wjd().

- Add guess() as another supporting function for wjd(), to produce 
  initial guesses of moles of species satisfying a given elemental
  bulk composition, and a Suggests dependency on 'limSolve'.

- New function i2A() for generating a stoichiometric matrix from indices
  of species in the thermodynamic database.

- Add protein.equil() for step-by-step calculation of chemical
  activities of proteins in metastable equilibrium.

- Add an objective function DGtr() for calculating the Gibbs energy of
  transformation of a system at constant temperature, pressure and
  chemical activities of basis species.

- Add msgout(), which is a modification of message() from base R. Now
  used instead of cat(), to allow suppressing messages (e.g. during
  testing with test_that).

CHANGES TO ARGUMENTS:

- In equilibrate(), the argument 'logact' (inherited from diagram()),
  specifying the logarithm of activity of the balanced quantity, has
  been renamed to 'loga.balance'. In the result, rename 'logact' to
  'loga.equil', containing the equilibrium logarithms of activities
  of the species of interest.

- The 'residue' argument of diagram() has been changed to 'normalize'
  in diagram() and equilibrate(). normalize=FALSE is always the default,
  including for systems of proteins.

- In balance(), the value of 'balance' used to indicate protein length
  has been changed from 'PBB' to 'length'.

- Everywhere it appeared, the logical argument 'do.plot' has been 
  renamed to 'plot.it' (diagram(), revisit(), findit(), transfer()).
  (This scheme is more consistent with e.g. qqplot().)

- 'do.phases' argument in subcrt() and affinity() has been renamed to
  'exceed.Ttr'. When that argument is FALSE (the default for subcrt()),
  the Gibbs energies of mineral phases at temperatures beyond their 
  transition temperature are set to NA, instead of 999999 used 
  previously.

- ZC() now accepts a numeric argument, referring to one or more species
  indices in the thermodynamic database.

CHANGES TO OUTPUT:

- subcrt() now only outputs T (temperature), P (pressure) and rho
  (density) columns if there is more than one T-P point ... makes
  unlist()ing the results easier (used in element.mu() and 
  basis.logact()).

- subcrt() now outputs NA values for properties at temperatures above
  the critical temperature of H2O, when Psat is being used.

- read.blast() accepts NA for 'similarity', 'evalue' and 'max.hits' 
  options. Descriptive column names are now assigned to the data frame
  returned by the function.

- energy.args() (called by affinity()) shows units in messages about
  limits of variables.

- EOSvar() has new variables invPPsi and invPPsiTTheta, used for
  temperature- and pressure-dependent regressions in the revised HKF
  equations of state.

- thermo.plot.new() now saves the graphics device parameters 
  (par(no.readonly=TRUE)) to thermo$opar the first time the function is 
  called, allowing the parameters to be restored after running examples
  that change them.

- Rewrite mod.obigt() (it's now used by info() when adding proteins) and
  add today() for returning today's date in the format used in SUPCRT 
  files.

REFACTORING OF FUNCTIONS:

- Split the primary functionality of makeup(), parsing of chemical 
  formulas, into a smaller makeup() function and supporting functions 
  count.charge(), count.formulas(), and count.elements(). The new 
  functions make extensive use of regular expressions, and no data 
  frames. Running makeup() over the ca. 3000 formulas in thermo$obigt 
  drops from ~35 to ~5 seconds on one machine.

- New function as.chemical.formula() to replace the previous
  functionality of makeup() for making string representations of
  chemical formulas.

- Replace element() with two separate functions, mass() and entropy().

- Units setting interface is now split between three separate functions:
  P.units(), T.units() and E.units().

- Replace describe() with describe.basis(), describe.property(),
  describe.reaction(). It is now fairly easy to make legends showing
  temperature, pressure and chemical activities with italic symbols,
  subscripts, and units.

- Reorganize axis.label() and its supporting functions 
  (now expr.species(), expr.property() and expr.units()). Add an example
  showing a plot annotated with chemical formulas and reactions.

- The functionality of info() is split into info(), info.character(),
  info.approx(), info.numeric() and info.text(). For ease of use, single
  approximate matches are accepted by info(), and searches for 'H2O' in
  the 'aq' state now return H2O(liq).

- The basis definition functionality of basis() is split into basis(), 
  is.basis(), put.basis(), mod.basis(), preset.basis() and 
  preset.logact().

- The basis swapping functionality of basis() is split into 
  basis.matrix(), element.mu(), basis.logact() and swap.basis(). Rename
  basis.comp() to species.basis(), and remove expand.formulas().

- Split aminoacids() into aminoacids() and count.aa().

- The monolithic protein() function no longer exists; it has been
  superseded by iprotein(), ip2aa(), get.aa(), dl.aa(), read.aa(),
  sum.aa(), and aa2eos(). The user shouldn't notice significant changes
  (other than in the composition of messages) when including proteins in
  functions like subcrt() and affinity().

- Functionality of get.protein() is split into more.aa() (amino acid
  compositions from model organisms) and stress() (proteins identified
  in stress response experiments).

- get.expr(), for reading abundances or expression levels of proteins
  from variously formatted data files, is renamed to read.expr(), with
  retrieval of amino acid compositions of proteins moved to more.aa().

- Replace ionize() with a completely rewritten and much easier to use
  ionize.aa() for calculating the additive ionization properties of 
  proteins; also add A.ionization(), usually invoked by affinity()

- residue.info() mostly replaced by protein.basis().

- Rename equil.react() to equil.reaction(), and give it parallel
  potential (via palply) and a more efficient algorithm for determining
  limits of the uniroot search. Also remove a redundant argument; 
  arguments are now identical to those of equil.boltzmann() (and that
  function was renamed from equil.boltz()).

- Objective functions used in revisit() and findit() now each have
  their own definitions, with an attribute indicating whether the
  function is minimized or maximized.

- Separate idealgas.IAPWS95() and residual.IAPWS95() from
  water.IAPWS95() (in order to write some test_that tests.)

DATA UPDATES:

- Add 148 liquid and 148 crystalline acyclic isoprenoids, polycyclic 
  alkanes, polynuclear aromatic hydrocarbons (PAH) and 62 crystalline
  double ether-bonded or ester-bonded n-alkanes from Tables 16-23 of 
  Richard and Helgeson, 1998. The properties of the following 
  crystalline and liquid compounds, taken from those tables, replace the
  previously entered values from the "reference model compound" tables
  appearing earlier in that paper: biphenyl, naphthalene, 
  1-methylnaphthalene, 1,8-dimethylnaphthalene, 2,3-dimethylnaphthalene
  (cr only), anthracene, pyrene.

- Add 13 crystalline, 29 liquid and 39 gaseous organic iodine compounds
  from Richard and Gaona, 2011.

- Add crystalline peptide sidechain and [AABB] and [PBB] groups and 
  dipeptide model compounds and revised equations of state parameters 
  for crystalline leucine and revised standard Gibbs energies of
  crystalline and aqueous methionine and its aqueous sidechain group
  [Met], from LaRowe and Dick, 2012. Also, use revised volumes of all
  other crystalline amino acids given in that paper together with those
  of Lys:HCl and His:HCl calculated using an effective volume of HCl(cr)
  equal to 33.8 cm3/mol (difference between Arg and Arg:HCl).

- Add 6 aqueous chloroethylenes from Haas and Shock, 1999.

- Move superseded aqueous methionine sidechain group [Met] to 
  OBIGT-2.csv so that its properties are available for reproduction
  of published results (relevant to some examples in the package).

- As revised makeup() function now strictly interprets a signed value at
  the end of a chemical formula as charge, the formula of the electron
  in thermo$obigt is changed from "Z-1" to "Z0-1".

- In thermo$obigt, change the value for standard Gibbs energy of H2O 
  from -56688 to -56687.711481 cal mol-1, to be consistent with the 
  value generated by the fortran code from SUPCRT92 (H2O92D.f). Although
  the latter is the source of the properties of H2O for many functions
  in the package, there is an occasional function that accesses the
  tabulated value at 25 degrees C (e.g., element.mu() and 
  basis.logact()).

- In protein.csv, change organism code BACST (Bacillus 
  stearothermophilus) to GEOSE (Geobacillus stearothermophilus) for SLAP
  and DPO1 proteins, and also apply changes in vignettes and examples.

- In stress.csv, change ECO to Eco and SGD to Sce.

EXTDATA UPDATES:

- In extdata/refseq, scripts and data files were updated for NCBI
  Reference Sequence (RefSeq) release 55 (2012-09-17).

- In extdata/bison, sample BLAST output files for Bison Pool metagenome
  use target database generated from RefSeq release 55.

- Add P(ressure) column to extdata/cpetc/SOJSH.csv and a stopifnot() 
  test for similarity to the experimental data to the example in 
  water.Rd.

- In extdata/protein rename ECO.csv.xz to Eco.csv.xz and SGD.csv.xz to
  Sce.csv.xz.

- In extdata/thermo add RH98_Table15.csv; this file together with new
  function RH2obigt() is used to calculate thermodynamic properties of
  organic compounds using group contributions from Richard and 
  Helgeson, 1998.

DOCUMENTATION AND VIGNETTES:

- Add a Known Bugs section to the package help page.

- Move 'extra' examples previously available in longex() to individual
  demos, and add a function demos() to run them all.

- All examples now run without any warnings (at least, as intended).

- Add an example based on Shock and Canovas, 2010 for the 'transect' 
  mode of affinity().

- ionize.aa() has examples of contour plots as a function of temperature
  and pH.

- Group residue.formula(), protein.formula(), protein.name(), 
  protein.length() into new documentation topic (util.protein.Rd).

- Add 'sideeffects.Rd' to document some of the side effects of 
  functions.

- Add 'objective.Rd' for the objective functions.

- To make output of examples reproducible, change mod.obigt() and 
  protein() to not include current date/time for new data entries 
  (remains an option for mod.obigt()).

- Move vignette sources to 'vignettes' directory.

- Remove vignettes formation.Rnw and xadditivity.Rnw 
  (and, for the latter, the package Suggests dependency on xtable).

- Rename protactive.Rnw to equilibrium.Rnw, with some changes.

- Add vignette wjd.Rnw to accompany the new function wjd().

- Add a release CHECKLIST to the installation directory of the package.

- Replace "degrees" with UTF-8 degree symbol in second argument of
  \eqn{}{}; add \encoding{UTF-8} to affected Rd files.

OTHER CHANGES:

- A matrix is now returned by makeup(), GHS() and some other functions
  to avoid the performace penalty associated with data frames.

- Add parallel as a Suggests dependency and replace mylapply() (based on
  'multicore') with palply(), as a wrapper for parLapply and lapply.
  palply() now invokes parLapply for length(X) > 10000 instead of 100.

- Comment out lines containing WRITE and STOP statements in src/H2O92D.f
  since they are discouraged by CRAN guidelines (and calls to them are
  unlikely to be encountered while using the package).

- Remove thermo$opt$verbose and thermo$opt$online.

- In c2s(), use 'collapse' argument of paste() instead of a for loop.

- Change R dependency to R >= 2.12.0, required for useDynLib in
  NAMESPACE to find the shared library on Windows.

CHANGES IN CHNOSZ 0.9-7 (2011-08-23)
------------------------------------

- Restore ZipData: no in DESCRIPTION, needed for building the package
  on R < 2.13.0 on Windows.

- Remove some incorrect titles for barplots in diagram().

- Add more informative titles for plots in first example of diagram.Rd.

- In EOSvar() and EOSlab(), name the first argument 'EXPR' in calls to 
  switch() to avoid partial matching with 'E' (expansivity).

- Shorten CHNOSZ-package.Rd and clean up formatting of some examples
  in other documentation topics.

CHANGES IN CHNOSZ 0.9-6 (2011-08-18)
------------------------------------

SIGNIFICANT CHANGES TO FUNCTIONS:

- New function browse.refs() for listing references for thermodynamic
  data and opening associated URLs. Supported by extdata/js/sorttable.js 
  for sorting table of references in browser.

- Add EOSregress(), EOSvar(), EOSlab(), EOSplot(), EOScoeffs() for 
  regressing and comparing equations-of-state parameters from 
  heat capacity and volume data for aqueous species.

- Add anim.plasma() and anim.carboxylase() for making animated GIFs of
  equilibrium activity diagrams. All anim.* functions now also work
  on Windows (with ImageMagick installed).

- New functions checkEOS(), checkGHS() for checking self-consistency of
  individual database entries. check.obigt() replaces 'check' argument 
  of info().

- New argument 'chains' for protein() can be used to specify the number
  of polypeptide chains in group additivity calculations.

- protein() now looks for protein backbone group named '[PBB]' for group
  additivity of crystalline proteins.

- Remove count.taxa(), splitting its functionality into read.blast()
  and id.blast(); new function write.blast().

- eqdata() can now also extract results for solids, saturation states of
  minerals, and major speciation of basis species from EQ6 output files.

DATA UPDATES:

- Rename 'source.csv' to 'refs.csv'. Add URLs for most references.

- Column names 'source1' and 'source2' in OBIGT.csv, changed to
  'ref1' and 'ref2'.

- In protein.csv, add amino acid compositions of 105 
  model proteins derived from metagenomic sequences at 
  Bison Pool (Dick and Shock, 2011).
  protein: overall, transferase, transport, dehydrogenase, ...;
  organism: bisonN, bisonS, bisonR, bisonQ, bisonP

- Add properties for aqueous bromine (Br2) and iodine (I2) from 
  Wagman et al., 1982.

- Add properties of crystalline and liquid groups from
  Helgeson et al., 1998 and Richard and Helgeson, 1998.

EXTDATA UPDATES:

- Reorganize the 'extdata' directory, putting all files into one
  of nine subdirectories (abundance, bison, cpetc, fasta, js, protein,
  refseq, taxonomy, thermo).

- To reduce installed package size, compress many data files in extdata 
  using xz.

- In extdata/bison, update partial protein BLAST output files for five 
  Bison Pool sites to use target database made up of microbial protein 
  sequences in NCBI Reference Sequence database version 47.

- In extdata/cpetc, new files 'Cp.CH4.HW97.csv' and 'V.CH4.HWM96.csv'
  contain experimental heat capacity and volume data for aqueous methane
  from Hnedkovsky and Wood, 1997 and Hnedkovsky et al., 1996.

- In extdata/refseq, rename 'taxid_phylum.csv' to 'taxid_names.csv'.

- In extdata/thermo/OBIGT-2.csv, fix incorrectly entered values of c1 from 
  Dalla-Betta and Schulte, 2010. Also apply corrected values of c1 and c2 
  and minor corrections for delta-H and S from Marini and Accornero, 2010.

DOCUMENTATION AND VIGNETTES:

- Change vignette file names and titles to be in the same alphabetical
  order to achieve the desired sorting in browseVignettes().

- Add .Rinstignore to inst/doc to exclude figure, bibliography and
  LyX files from the installation directory.

- Adapted 'formation.Rnw' to work with pdflatex, and remove
  Makefile in inst/doc.

- Add 'hs-chemistry.Rnw' with calculations of relative stabilities 
  of model proteins in a hot spring (Dick and Shock, 2011).

- Add 'protactiv.Rnw' with detailed information on calculating equilibrium
  activities of proteins, and comparisons with experimental abundances in 
  human blood plasma and E. coli.

- Add 'xadditivity.Rnw' with examples of group contribution calculations
  (using qr.solve()) and comparison of two group contribution schemes.

OTHER CHANGES:

- In DESCRIPTION, change Depends to a more recent R (>= 2.10.0),
  required for reading of compressed (including .xz) files.

- Add Suggests: xtable in DESCRIPTION (used for 'xadditivity' vignette).

- In DESCRIPTION, change "ZipData" to "BuildResaveData", keeping 
  the setting at "no". Resaving data as compressed files reduces the
  installed package size by about 200Kb, but complicates showing the file 
  formats to new users.

- Loading of shared object for calculating properties of H2O
  now done by useDynLib in NAMESPACE.

- Change .First.lib() to .onAttach(). In .onAttach, use 
  packageStartupMessage() instead of cat().
