Lazy access to all synapses in a local snapshot

Returns a lazy dbplyr table of every synapse in a local snapshot (see crant_use_snapshot()) for use with dplyr verbs. Nothing is read until you dplyr::collect() the result. This needs a local snapshot; there is no CAVE equivalent.

Usage

crant_synapse_data(
  side = NULL,
  details = FALSE,
  snapshot = getOption("crantr.synapse_snapshot", "latest"),
  root = crant_snapshot_root(),
  timestamp = NULL
)

Arguments

side

"post" to read the copy sorted by post_root when the snapshot has one. NULL (the default) and "pre" read the copy sorted by pre_root.

details

Whether to add supervoxel, position and size columns.

snapshot

The snapshot tag, or "latest" for the most recent snapshot in root. Defaults to the one chosen by crant_use_snapshot().

root

The snapshot folder. Defaults to crant_snapshot_root().

timestamp

A time (including "now") to get the synapses at, in place of snapshot.

Value

A lazy tbl.

Details

The table has columns id, pre_root and post_root (as integer64). With details = TRUE it also has pre_sv, post_sv, pre_x...post_z (positions of the pre and postsynaptic points in nm) and size.

Rows are sorted by pre_root, so filtering on a few pre_root values lets DuckDB skip most of the file. Use bit64::as.integer64() for root ids in filters.

With timestamp, the synapses are those of the newest snapshot at or before that time when it is at most getOption("crantr.synapse_max_age", 60) seconds older. Otherwise the edits made since the newest published snapshot are looked up in CAVE (seconds to minutes) and kept in memory for the rest of the session. Nothing is written to disk; use crant_update_snapshot() to keep a checkpoint for later sessions.

See also

crant_use_snapshot()

Examples

if (FALSE) { # \dontrun{
library(dplyr)
id <- bit64::as.integer64(crant_latestid("576460752684030043"))
crant_synapse_data("post") %>%
  filter(post_root == id) %>%
  count(pre_root, sort = TRUE) %>%
  collect()
} # }