-
-
we need to ensure all NBLAST queries can be added (different target settings) as well as any future R requirements, Hence I suggest the following changes:
https://github.com/openworm/org.geppetto.d…
-
I've been very happy with OrthoFinder so far. I'm attempting to resume a gene tree reconciliation from a result set that took ~2 weeks to compute. I have two ignored species 2: FRAX02, 29:FRAX34 whi…
-
Testing NBLAST on VFB2.0 (using Fru-F-500279).
Search for similar neurons worked fine.
GAL4 search just keeps thinking (4 attempts >10 min wait) and I don't get any results.
The GAL4 search on th…
-
Should VFB be using MEAN scores for NBLAST queries by default?
@jefferis Which should we be using as currently uses non-mean?
-
Working live Dataset Load:
```
{"id":"vfb:class:http://virualflybrain.org/data/Yu2013","iri":"http://virualflybrain.org/data/Yu2013","short_form":"Yu2013","shortform_autosuggest":["Yu2013"],"obo_id"…
-
A minimal example:
`nl = pymaid.get_neuron('annotation:glomerulus DA1')`
`nl.resample(1000, inplace=True)`
`res = rmaid.nblast_allbyall( nl )`
Additionally, my understanding is, since the nblast…
-
```
id score
1 0.5375
2 0.5124
3 0.4930
4 0.4894
5 0.4648
6 0.4431
7 0.4423
8 0.4393
9 0.4230
10 0.4226
```
-
Submitter: NBlast
Email: nblastmax@gmail.com
There is a bug in coding of SuzyG enemy. It spawns two bullets, but they go in the same direction instead in opposite sides.
Code makes sense at first…
-
needed for VFB2.0 NBLAST support