Input specifications and emerging standards
Emerging standards for cell morphology, MorphML, and ion channels, ChannelML, are partially supported for the specification of models for PSICS. In general, most MorphML morphologies can be used as they are, and the majority of ChannelML files for voltage gated channels are compatible subject to the assumption of a default single-channel conductance.
More details of the current status of NeorML imports in PSICS and suggestions for ways to improve compatibility can be found in a spearate section: Importing NeuroML.
ChannelML files that specify models with concentration dependent gating, as in calcium activated potassium channels, cannot be used since PSICS cannot currently compute such models.
Some ChannelML files can express models that can be computed by PSICS but that cannot be imported automatically. These are mainly HH-style models with extended or ad-hoc specifications of rate equations beyond the standard alpha/beta (forward rate, reverse rate) or tau/inf (time constant and eqilibrium open fraction) forms. Such models generally can be computed by PSICS, since it supports arbitrary coded transitions, but the conversion is not automatic. It is likely to be more effective to convert the relatively small number of such models by hand, rather than developing a general solution for importing them.
The kinetic scheme structures in ChannelML are relatively close to those used in PSICS, and the standard fitted transition types (sigmoid, exponential, etc) can be imported. The default biophysical transition parameterization used in PSICS is not specifically supported although it can be expressed as a generic equation. Since any voltage-gated channel models can be expressed as compound kinetic schemes, the kinetic scheme elements (ks_gate) of ChannelML provide the cleanest export route for PSICS models, although they are note widely supported by other systems yet.