neuprint_connection_table returns different values as on neuprint website

[istvantaisz]

Synapse numbers retrieved are different from the ones found on https://neuprint.janelia.org/ for neuron 580227374. Input: 5176 vs 4908, Output: 2133 vs 6498

library(neuprintr)
M_inputs  = neuprint_connection_table(580227374, prepost = c("PRE"))
nrow(M_inputs)
[1] 854
sum(M_inputs$weight)
[1] 4908
M_outputs  = neuprint_connection_table(580227374, prepost = c("POST"))
nrow(M_outputs)
[1] 1368
sum(M_outputs$weight)
[1] 6498

My dataset version matches the online one: neuprint_dataset = "hemibrain:v1.0.1"

Related to #63 and #67

[romainFr]

Hi @istvantaisz ,

The numbers reported on the website are the total numbers of pre and post synaptic sites respectively. In neuprintr, those would be the pre and post columns returned by neuprint_get_meta. neuprint_connection_table differs in two ways:

• by default, it only returns connections to neurons (and not unidentified fragments). This is controlled by the all_segments option. This explains the first discrepancy you see
• it reports the weights, which in neuprint are always correspond to the number of postsynaptic sites between two neurons. That explains the second discrepancy

In other words:

• neuprint_connection_table(580227374, prepost = c("PRE"),all_segments=TRUE) will sum to the same number as the #post column on the neuprint website
• neuprint_connection_table cannot sum to the #pre column because it is reporting weights, not number of presynaptic sites (the same way that on the neuprint website, were you to sum the #connections column, you wouldn't get the #pre column of the parent neuron)

Let me know if that helps, and thanks for reporting!

[istvantaisz]

Hi Romain, thanks for your response. I missed the all_segments option.

And indeed when I exported all outputs I got a third number for all outputs, which matches with nrows from neuprint_connection_table(580227374, prepost = c("POST"), all_segments = T)