Annotation Files

Although Knossos can still load and export *.nml files, the recommended way is to use the new annotation files.

The annotation files *.k.zip are zip archives that can currently contain 3 things:

• A file annotation.xml which is formerly known as *.nml containing general information about the whole annotation and optionally a skeleton (for an easy start the contents are currently the same as you’d get when exporting as *.nml, but the vision is to split the skeleton from skeleton independent information in the future)
• several *.seg.sz files containing segmentation overlay cubes which are loaded at higher priority than files from disk and saved when cubes got edited
  the filenames consist of: the experiment name, '_', the magnification, 'x' 'y' and 'z' with their respective cubes coordinates
  e.g. 2012-03-07_AreaX14_mag1 _ mag1 x41 y40 z22 .seg.sz
• A file mergelist.txt containing information about objects which span multiple overlay ids with the format specified below

###Merge list

The merge list is used to group supervoxels (groups of voxels with the same unique ID) into objects and assign information to them. It’s just a text file where each object occupies 4 lines.

Object = Object_ID, immutability_flag, supervoxel_IDs, '\n', position, '\n', category, '\n', comment;

Object_ID = uint64, ' ';
immutability_flag = (0|1), ' ';
supervoxel_IDs = uint64, {' ', uint64};
position = int32, ' ', int32, ' ', int32;
category = comment = −’\n’;

The immutability_flag is used to represent if an object should persist in the list after it was merged by proofreaders (i. e. for later unmerging, usually all initial objects given to proofreaders are marked as immutable). category and comment each occupy a line on their own – without any escape characters.

Example:

3150 1 5352 3297 3971 4104 2654 4153 4046 2257 3624 3150 4100
5333 5267 2880
mito
whatever one may want to add here
1 1 1 2

test
1024 2048 4096
3151 0 2257 2654 2663 3018 3023 3029 3150 3294 3297 3298 3363
neuron
proofread merge