The format for a sif file is:
nodeID1 [tab] [PROMOTES | REPRESSES | REGULATES] [tab] nodeID2
where the link sign (positive, negative, or neutral), is denoted
in the middle interaction field.
In the top-level "Full Genome" root network, the nodeID is
either just the gene name (which is globally unique), or in
the case of non-gene nodes, the node name followed by a
period, followed by the unique internal node ID. If the
node is unnamed, then the name is replaced by the node
type inside parenthesis. For example:
(bubble).15 [tab] PROMOTES [tab] Delta
nB-TCF.25 [tab] REPRESSES [tab] Wnt8
Below the top level, regions are present, and the region
name, in square brackets, followed by a colon, is prepended
to the node ID. Note that region names are unique.
Also, for non-genes, the node instance number is appended
with a colon. For example:
[PMC]:(bubble).15:0 [tab] PROMOTES [tab] [PMC]:Delta
[Mesoderm]:nB-TCF.25:4 [tab] REPRESSES [tab] [Mesoderm]:Wnt8
Note that although, in almost every case, this method should
create unique IDs, it is possible to name a node that collides.
If the system detects such a collision, it will prepend "_"
underscores to the ID as needed to make it unique. This test
for uniqueness ignored blanks in the names and is also
case-insensitive.
Next, if the node activity level is not the default
"active", the system will add an activity suffix
("inactive", "vestigial", or "variable") to the node ID,
contained in curly brackets "{}".
Finally, in dynamic (e.g. time-slider) models, you can add "extra nodes"
to the model that are not in one of the included regions. This
is often done to show, for example, an incoming signal node in
a neighboring region, or the link inbound from a neighboring region.
In these cases, the region name will the replaced by the string
"--no region--", followed by the node ID and instance ID, e.g.:
[--no region--28:1]:(bare).28:1{inactive}
Any singleton nodes in the network will be appended to the
end in a single-field node-only interaction line.
-- Bill