BANC

library(fancr)
library(dplyr)

Introduction

The BANC is a female brain and nerve cord dataset imaged by Wei Lee’s lab (Minsu Kim and Jasper Phelps). See its wiki. Since the BANC infrastructure is most similar to FANC (the female adult nerve cord), we have decided to use the fancr package to provide some basic functionality to interact with the BANC dataset.

with_banc and choose_banc

Like the underlying fafbseg package, fancr has a concept of an active segmentation.

You can now use choose_banc to set the active segmentation to BANC, which should target most fanc_ functions to use the BANC. Of course if you want to use a different dataset (fanc, flywire) you need to remember to switch back.

A potentially less intrusive approach uses with_banc() to temporarily switch the default segmentation to the BANC, run your code, and then switch back again. Just wrap flywire_.* or fanc_.* functions with this to target them at the BANC.

# supervoxel id to root id
with_banc(fafbseg::flywire_rootid('76071705504180616'))
#> [1] "720575941593486806"

is this id up to date?

with_banc(fafbseg::flywire_islatest('720575941472355131'))
#> [1] FALSE

Find up to date root id

with_banc(fafbseg::flywire_latestid('720575941472355131'))
#> [1] "720575941593486806"

CAVE tables

cell_info=banc_cave_query('cell_info')
cell_info %>%
  filter(tag2=='anterior-posterior projection pattern') %>%
  count(tag)
#> # A tibble: 3 × 2
#>   tag                n
#>   <chr>          <int>
#> 1 ascending       2190
#> 2 descending      1267
#> 3 intersegmental    16

Or we can just request a specific tag type

nc=banc_cave_query('cell_info', 
                   filter_in_dict = list(tag2='anterior-posterior projection pattern'))
nrow(nc)
#> [1] 3473

L2 skeletons and dotprops

BANC is already set up with an L2 cache which can be used to get rapid neuron skeletons.

n=with_banc(fanc_read_l2skel("720575941552483585"))
#> Default dataset set to "brain_and_nerve_cord".
plot(n)

Similarly we can make “dotprops” objects suitable for NBLAST.

ids <- with_banc(flywire_rootid(c("76070056370827492", "76142074315307608", "76142074315367069")))
dps=with_banc(read_l2dp(ids))
#> Default dataset set to "brain_and_nerve_cord".
plot(dps)

library(nat.nblast)
aba=nblast_allbyall(dps)
hc=nhclust(scoremat = aba)
#> The "ward" method has been renamed to "ward.D"; note new "ward.D2"
plot(hc)

Using multiple datasets

If you use choose_banc() to set BANC as the default dataset, then don’t forget to use choose_fanc() or fafbseg::choose_segmentation() to switch to a different active segmentation.

If you expect to switch between multiple datasets in a session, we recommend avoiding the choose_banc() function since this will cause all fanc_.* functions to target BANC. Instead you can use with_banc()

with_banc(fanc_rootid(c("76070056370827492", "76142074315307608", "76142074315367069")))

This does mean a bit more typing but it will probably avoid problems.

Note that you do opti to use choose_banc(), then you can use with_fanc to ensure that any fanc function indeed targets FANC.

choose_banc()
# a banc supervoxel id
fanc_rootid("76070056370827492")
#> [1] "720575941494603975"
# a fanc supervoxel id
with_fanc(fanc_rootid("73186243730767724"))
#> [1] "648518346499897667"