Tubemap Hacking Roadmap

graph-genome / MatrixTubeMap

Using a Matrix to show coverage of multiple genomic sequences using a prototype hack of Sequence Tubemap

MIT License

7 stars 0 forks source link

Tubemap Hacking Roadmap #1

Closed josiahseaman closed 4 years ago

josiahseaman commented 4 years ago

[x] Grab example data state
[x] Change substitution text
[x] Speed up development build / edit cycle
[x] Introduce "link" mismatch type into code
- [x] has attr() query: "javascript to execute"
[x] Trigger JavaScript on link click
[x] Nodes are transparent
[x] Verified control over Read ordering
[x] Verified control over Node ordering

Bugs to Squash

[x] Link Position not right on all lines
- Binary search?
[x] Collapsing multiple read into one line with gaps
[x] Two orange reads overlapping - start 0 length 1944

josiahseaman commented 4 years ago

control over Node ordering

id does not control ordering
id controls coloring (indirectly) trackColor = forwardReadColors[track.id % forwardReadColors.length];
Sorting controlled by order a read is in the list, not id. There's an auto-densify you need to be cautious of with short reads.
finalNodeCoverLength is misleading. This is the stop index when it is the same node. If stop < start, it will still be rendered and possibly quite long.
editing Node id seems to have no effect now, only Node order in list?
It's important finalNodeCoverLength is not longer than the Node it's in or there is a graphical bug
Declaration order of Nodes doesn't matter now as long as "tracks" sequence order matches the intended order reads list node names in.

josiahseaman commented 4 years ago

finalNodeCoverLength was changed to absolute coordinates inside of one node. e665f5c979509b846350e5b7dc00b16722f78216

josiahseaman commented 4 years ago

As a generalization, links tend to apply to a full column. This makes Erik's connecting lines at the bottom more preferable.

josiahseaman commented 4 years ago

Yeast dataset:

Error stripes are more prominent. First priority will be removing those. DRB1 Errors: Tracks 64 & 107. Errors come guessing the total length of the Node wrong. This makes reads whose end is after the node end. Solution: make the node longer. This comes from --bins= param in tubeify.

josiahseaman commented 4 years ago

This visualization seems pretty good at hilghlighting the sections that are linear, the "hubs" that are irreconcilable snarls, and (when dense) the nodes that are mutually exclusive.