I’m calling the sparse LAP tracking framework from another framework, I get encouraging results except that it seems spots that were not linked during the FTF linking step won’t be linked during the Segment linking step (I guess because they are not present in the graph used for the segment linker). Is there a way to include those spots in the Segment linking step (as segment ends & segment starts at the same time) ?
All the LAP trackers in TrackMate implements the framework of Khulud Jaqaman (a stripped version). Unfortunately, if a spot is not considered in the FTF phase, it will not be considered in the segment linking phase.
If you want to change this behavior, you have to write your own tracker, that can derive from this one.
Thank you for your quick reply. In my case, this behavior led to a lot of data loss.
I found an easy workaround: Before the segment linking phase, I add clones of the unlinked spots to the graph, and connect the originals with theirs clones.
I have worked on a similar approach, I will be happy to share ideas.
You don’t need to clone them if you are manually editing tracks.
If you link a track to a single spot that is not visible in TrackScheme, it will become visible iN TrackScheme automatically.
Actually the workaround I described just allows to add the unlinked spots from the FTF phase in the graph, so that they are automatically recognized as “segment ends” (cloned spots) and “segment starts” (original spots) during the Segment linking phase. After the segment linking phase I remove the cloned spots and merge the links. Spots need to be cloned because a spot cannot be linked with itself in this kind of graph.