Forums

Greetings,

Joerg Weimar posted a question on May 16 about the units of
initial values in SBML. His message was a follow-up to a
topic of discussion during the 7th Forum meeting last week.

The general sense from the meeting is that people agreed
that the default units used for the 'initialAmount' on
species does not reflect the actual use of the field in
practice. SBML defines the units on 'initialAmount' as
being the built-in 'substance' unit, but instead, most
models today appear to have been written with the assumption
that the units are actually 'substance'/'volume'.

Several possible solutions to the problem present
themselves:

a) Keep the default units of 'initialAmount' as 'substance'
as it is defined currently. Most software and models
would need to change to comply with the standard as it is
written.

b) Introduce a new field, 'initialConcentration', with units
of 'substance'/'volume', and make 'initialAmount' and
'initialConcentration' be mutually exclusive.

c) Change the default units of 'initialAmount' to be
'substance/volume'. This would reflect the most common
usage today.

The text of Joerg's message is appended below; he provides
some additional analyses of the implications of the
different options.

Please reflect on and discuss these issues on the list. The
voting system will be enabled beginning this Friday and we
will post further information at that time.

If you have questions about this topic or the reason for the
vote, please feel free to call Mike or Andrew directly.

-- Mike & Andrew

> From: "Joerg R. Weimar" <J.Weimar@tu-bs.de>
> To: sbml-discuss@caltech.edu
> Subject: [sbml-discuss] Initial conditions
> Date: Fri, 16 May 2003 13:00:39 +0200
>
> Issue with SBML L2: definition of initial values.
>
> Currently the standard defines initial conditions for species in terms
> of units of substance. Units of substance can be "mole" or "item". This
> is different from concentrations. On the other hand, the species symbol
> as used in rate laws refers to concentrations. Therefore the initial
> assignment uses different units (and necessarily different numbers,
> unless the compartment volume is 1) from the use of the symbol. Under
> the current definition, if the volume of a compartment is changed, all
> initial values must be changed, if the intention is to keep the same
> concentrations. Currently not all tools abide by the standard, some
> tools assume that "initialAmount" is actually a concentration.
>
> Possible solutions:
>
> 1. Enforce the current standard and live with the disadvantages.
>
> Advantages:
> * The standard is not changed.
> Disadvantages:
> * If the volume of a compartment is changed, all initial values must
> change.
> * The value used for initial assignment is different in value and units
> from the value returned for this symbol when used in formulas.
>
>
>
> 2. Introduce a second possibility for initialization using
> concentrations. One way would be to have for Species two attributes
>
> initialAmount : double {use="optional"}
> initialConcentration: double {use="optional"}
>
> with the specification that "initialAmount" and "initialConcentration"
> are mutually exclusive. The units would still refer to the units of
> substance, for which there are the options "mole", "item", and units
> derived from these, such as "10^-3 mole".
>
> Advantages:
> * Each tool could use the initialization method it prefers. Tools that
> currently assume concentrations only have two small changes: write
> "initialConcentration", when reading check which attribuute is present
> and scale if "initialAmount is present".
> * Existing models that abide by the standard are still valid.
> * Existing models that assumed concentrations can be fixed with a simple
> text replacement operation.
>
> Disadvantage:
> * Tools have to check for presence of two attributes for setting the
> initial values.
>
>
>
> 3. Interpret the units to figure out whether they are of type substance
> or of type substance/volume and rescale accordingly when reading in a
> model. The default units are defined in the standard and imply that
> without specification, units of substance are used, which are "mole" and
> not "mole/liter".
>
> Advantages:
> * No changes to attributes.
> * Existing models that are correct (use units correctly or write numbers
> which have the correct units as specified for the default units) remain
> correct.
> * extensions to membranes (with area instead of volume) can also use this
> mechanism.
>
> Disadvantages:
> * Currently no tool parses and analyses units.
> * the name initialAmount would be misleading if it can also refer to
> concentrations.
>
> --
> PD Dr. Joerg R. Weimar, Inst. f. Wissensch. Rechnen, TU-Braunschweig
> J.Weimar@tu-bs.de, http://www.tu-bs.de/institute/WiR/weimar
> Tel. +49-531-391-3006 Mail: D-38092 Braunschweig