More uniform fields

[jefferis]

It would be useful to add more fields that are uniformly available across datasets. The following

• sex (obvious from the dataset, but I think better to have directly returned)
• organ (or tissue or division or some such).
  • Could be brain or VNC.
  • Should it be based on somatic origin? or synapse overlap?
  • If synapse overlap, we need to represent both: CNS?
  • how about sensory neurons: same?

[jefferis]

I think this should be soma based e.g. soma_division and there should be options for e.g. head and body to handle sensory neurons. But body strictly refers to the whole thing, so we might need to say thorax+abdomen which is a bit unwieldy.

[jefferis]

That said, for CNS datasets it would frequently be helpful to resolve connectivity into brain vs VNC. So soma_division works well for metadata, but for cf_partners() some kind of synapse_location should perhaps be returned.

[jefferis]

@mmc46 what do you currently do to get a brain vs vnc flag for malecns?

[mmc46]

I mostly use 2 things:

• filtering by superclass and
• get bodies in VNC via neuprintr::neuprint_bodies_in_ROI
• set some exceptions manually, excluding central neurons that extend a bit down the neck or the neck motor neurons whihc I don't want to include as invnc.

I include as invnc the superclass descending_neuron and other descending ones, though logically if they have somas in the brain they probably should not be TRUE for invnc.

load_mcns_dvid <- function() {
  
my_vnc_superclasses <- c("ascending_neuron", "ascending_neuron_tbc", "descending_neuron", "descending_neuron_tbc", "intrinsic_neuron", "intrinsic_neuron_tbc", "sensory_ascending", "sensory_ascending_tbc", "sensory_descending" ,"vnc_tbc", "efferent_descending", "efferent_descending_tbc", "efferent_ascending", "efferent_ascending_tbc")
  
  mdap = mcns_dvid_annotations()
  
  
    conn_mcns = malecns::mcns_neuprint()
    
    vnc_bodies = neuprintr::neuprint_bodies_in_ROI(roi = "VNC", all_segments = F, conn = conn_mcns)
    vnc_bodies$bodyid <- as.character(vnc_bodies$bodyid)
    
  mda <- mdap %>%
    mutate(across(where(is.character), ~ if_else(. %in% c("", "NA"), NA_character_, .))) %>%
    mutate(across(c(bodyid, group, manc_bodyid, manc_group), as.character),
           invnc = case_when(
             superclass %in% my_vnc_superclasses ~ TRUE,
             bodyid %in% vnc_bodies$bodyid & superclass == "central" ~ FALSE,
             bodyid %in% vnc_bodies$bodyid & (superclass == "motor" & str_detect(exit_nerve, "CvN")) ~ FALSE,
             bodyid %in% vnc_bodies$bodyid ~ TRUE,
             TRUE ~ FALSE
           )) 
  
  return(mda)
}

[mmc46]

Even with the above I can see there are some neurons that fall through the cracks. So I'm getting 16 sensory neurons in the VNC that don't get the invnc. They don't have any synapses so I assume that's why they don't get picked up by neuprintr::neuprint_bodies_in_ROI.