Dataset Open Access

SPARC Connectivity Knowledge base of the Autonomic Nervous System

Tom Gillespie; Bernard de Bono; Monique Surles-Zeigler; Natallia Kokash; Fahim Imam; Susan Tappan; Jyl Boline; Jeffrey Grethe; Maryann Martone

Data curator(s)
Joseph Menke

The SPARC Knowledge base of the Autonomic Nervous System (SCKAN) is an integrated graph database composed of three parts: the SPARC dataset metadata graph, ApiNATOMY and NPO models of connectivity, and the larger ontology used  by SPARC which is a combination of the NIF-Ontology and community ontologies.

The fastest way to get querying is to follow the instructions in the SCKAN readme file.

For background information please see and the SPARC portal resource page about SCKAN.

This release contains the raw and compiled data for SCKAN. The release-*.zip contains raw data inputs along with the Blazegraph journal file, the sparc-sckan-graph-*.zip contains the SciGraph database, and sckan-data-*.tar.gz is a Docker image that contains the Blazegraph journal file and the SciGraph database along with the configuration files for running each of the servers. The image is intended  to be used as a data volume with another Docker container that runs the SciGraph and Blazegraph server software.

The Docker image containing this data is available live and is likely easier to use than the archived image included in this release. See the SCKAN readme file for the most up-to-date instructions.

We would like to thank the members of the SAWG (SPARC Anatomy Working Group, RRID:SCR_018709) for their work on the various connectivity models included in this release.

This work was funded by the NIH Common Fund under 3OT2OD030541-01S1.

There are two known issues with this release affecting aacar-6 and sdcol-f. See and They will be resolved in the next release. Here is an overview of the general changes in the 2022-08-19 release of SCKAN reproduced from in general there have been many changes made in order to bring the apinatomy model representations in line with their respective diagrams (i.e. docs/{model-name}.svg changes made for general consistency 1. all ontologyTerms for anatomical parts have been updated to use ILX: or UBERON: terms, use of any other prefix is a bug and should be reported as such 2. processes passing through white and gray rami also pass through the corresponding sympathetic chain ganglion 3. labels for dend-{bag,tube} have been update to reflect that the dendrite lyph templates are currently used in a way that is ambiguous to refer to both traditional dendrites and structures that are axons with sensory endings, a future update will disambiguate these cases and also address issues in the peripheral nervous system where an absense of explicit dendrites was incorrectly interpreted as the presenece of a dendrite (the default assumption is different from the central nervous system) 4. in most, but not all cases we have switched to using levelTargets to model neuron populations with somas in multiple locations that send processes through the same structures e.g. kblad 4/8, aacar 5 5. in cases where a housing region was unknown we selected something like a least upper bound to constrain the location e.g. for axons projecting into spinal white matter from the periphery we listed central nervous system as a target, and for sensory processes whose anatomical original was unknown we listed multicellular organis aka whole body 6. where more specific ontology terms are available for specific lyphs and layers we attempt to use them e.g. myenteric plexus of stomach instead of the more general myenteric plexus individual changes are too numerous to list but nearly all have been documented specifically as part of the qc process, ping for details if needed, some general patterns for issues that were correct follow 1. inversion of layer ordering, e.g. lyphs specified inside out so that e.g. the colon was inside out with the serosa as the inner layer and the lumen the outer layer, or e.g. blood specified as being outside the blood vessel wall 2. frame shifts, insertions, etc. corresponding to biological genomic mutations, induced by copy/paste errors of various kinds, nearly all leading to off by 1 errors or incorrect aliasing, or sometimes incorrect typing of neurons with similar structure 3. axon/dendrite confusion independent of axon sensory/dendrite issues there are a number of cases where the notation used in apinatomy diagrams can be ambiguous, some general categories where a mismatch should be expected are as follows 1. branching of neurons into the same structure, in nearly all cases where a diagram represents multiple lyphs that pass through or terminate in a single housing lyph only a single branch is represented, e.g. sstom {11..14}, aacar {4,6}, in many of these cases the branching is induced by projections onto multiple different cell populations and could be handled during model expansion 2. neurons passing into and out of an internal lyph back into the parent, if a branch happens inside the internal lyph the notation is equivalent to a branch occuring in the housing lyph prior to entering or even after exiting the internal lyph, e.g. pancr {2,4} 3. the location of implicit nodes in cases where continuous lines in a diagram could be in one or more locations (related to the internal vs housing lyph boundaries issue mentioned above), see e.g. aacar {4,6,11,13}, digrams can and do vary greatly from models in such cases notational confusion in apinatomy diagrams 1. orthognoal crossing lines of the same color that are numbered but otherwise appear continuous, we have fixed most of these cases, but every single person who has ever looked at the kblad diagram has confused populations 2 and 4 (hopefully the recent changes are sufficient to disambiguate the two), this issue is not unique to apinatomy diagram notation, anyone reading a diagram should still take care to check whether additional notation exists when an orthognal line crossing of lines is encountered 2. implicit dendrites, the default assumption switches across the cns/pns boundary 3. lack of indication of the axis of rotation for lyphs, this is hypothesized to be the primary cause of the layer order inversion issues that we have encountered, some form of asymmetry is critical for quick an accurate reading of diagrams, cysts and bags can usually be read without issue due to their inherent asymmetry, but tubes need explicit axis, readers encountering a tube without specification should take special care and also report the issue 4. use of a scaffold region by itself without a corresponding lyph from a connectivity model, most of these have been fixed, but some cases may not have been corrected, this is particularly confusing when the region is empty and used as a reminder that there is something in there that a process passes through, but where the exact location is unknown, in such cases the region should be left out, but we often instinctively want to close the world and include it in the connectivity diagram, thus explicit lyphs in such regions are critical for clear communication 5. implicit boundaries between regions, e.g. kblad {6,7}, bromo 1, sdcol b, these have been almost entirely resolved in new diagrams by keeping even regions that are contiguous separate 6. the converse of 5 seen in e.g. pancr 4 where proximal duodenum and gastric antrum are intended to be continuous but there is not an explicit link/node that denotes this in the diagram 7. chains drawn as going directly to an intermediate layer of a lyph without explicitly passing through the overlying/underlying layers, in all cases the chain must pass through the overlying/underlying lyph, but there is not bottom up way to know whether the outer or inner layer is the entry point, in most existing cases e.g. kblad {2,3} it is the outer layer another general trend that we have observed is that nearly all errors that we found in the ApiNATOMY models after initial checking against NPO were things that NPO could never have caught because they were e.g. incorrect identifiers for layers that the current version of NPO does not encode, the lyph layer inversions that literally no one caught because most tools do not care whether a sequece of housing lyphs is provided from inside to outside or outside to inside so long as their identifiers appear in the expected order (invariance to topology is an interesting blindspot in existing systems)
Files (1.1 GB)
Name Size
532.9 MB Download
107.0 MB Download
439.4 MB Download
All versions This version
Views 62463
Downloads 264
Data volume 7.3 GB1.2 GB
Unique views 47047
Unique downloads 172


Cite as