DESCRIPTION

@@ -1,7 +1,7 @@
 Package: coconatfly
 Type: Package
 Title: Comparative Connectomics of Public and In Progress Drosophila Datasets
-Version: 0.2.2
+Version: 0.2.3.9000
 Authors@R:
     c(person(given = "Gregory",
            family = "Jefferis",

NEWS.md

@@ -1,3 +1,19 @@
+# coconatfly (development version)
+
+* warn but still complete without errors when there are no partners at all for
+  inputs or ouptuts in `cf_partners()`. An example of when this happens is when
+  running `cf_cosine_plot()` for sensory neurons where all the inputs are below
+  threshold. #36
+
+# coconatfly 0.2.3
+
+* empty query fixes for all keys and banc metadata by @jefferis in https://github.com/natverse/coconatfly/pull/34
+* Fix malecns class in `cf_partners()` by @jefferis in https://github.com/natverse/coconatfly/pull/35
+* also support for rootSide in malecns in #35
+
+**Full Changelog**: https://github.com/natverse/coconatfly/compare/v0.2.2...v0.2.3
+
+
 # coconatfly 0.2.2
 
 * support malecns superclass/class/subclass change. To stay backward compatible 

R/datasets.R

@@ -23,6 +23,7 @@ match_datasets <- function(ds) {
 #' @examples
 #' abbreviate_datasets(c("flywire", "flywire", "hemibrain"))
 abbreviate_datasets <- function(ds) {
+  if(length(ds)<1) return(ds)
   ds=match_datasets(ds)
   abbrevlist=c(hemibrain='hb', flywire='fw', manc='mv', fanc='fv', malecns='mc',
                opticlobe='ol', banc='bc')

R/meta.R

@@ -136,6 +136,10 @@ malecns_meta <- function(ids, ...) {
   tres=malecns::mcns_neuprint_meta(ids)
   tres <- tres %>%
     mutate(side=malecns::mcns_soma_side(.)) %>%
+    mutate(side=case_when(
+      is.na(side) ~ rootSide,
+      T ~ side
+    )) %>%
     mutate(pgroup=malecns::mcns_predict_group(.)) %>%
     mutate(ptype=malecns::mcns_predict_type(.)) %>%
     rename(otype=type, type=ptype, ogroup=group, group=pgroup) %>%

R/partners.R

@@ -100,15 +100,20 @@ cf_partners <- function(ids, threshold=1L, partners=c("inputs", "outputs"),
                  chunk = neuprint.chunksize)
       tres=do.call(malecns::mcns_connection_table, c(args, ma1))
       # nb the type information we care about here is for partners
-      tres2=tres %>% dplyr::select(partner, type, name) %>% dplyr::rename(bodyid=partner)
+      tres2=tres %>%
+        dplyr::select(partner, type, name) %>%
+        dplyr::rename(bodyid=partner)
+
       tres$type <- do.call(malecns::mcns_predict_type,
                            c(list(ids=tres2), ma2))
       # set the soma side either from manually reviewed data
       tres <-  tres %>%
         dplyr::mutate(side=dplyr::case_when(
           !is.na(somaSide) & somaSide!='NA' & somaSide!='' ~ somaSide,
           T ~ malecns::mcns_soma_side(., method = "instance")
-        ))
+        )) |>
+        rename(class=superclass)
+
     } else if (n=='fanc') {
       args=list(fanc_ids(ids[[n]]),
                 partners = partners,
@@ -138,12 +143,22 @@ cf_partners <- function(ids, threshold=1L, partners=c("inputs", "outputs"),
         ))
     }
     tres=coconat:::standardise_partner_summary(tres)
-    tres$dataset=n
+    if(nrow(tres)>0) {
+      tres$dataset=n
+    } else {
+      tres$dataset=character()
+      warning("no ", partners, " found for `", n, "` dataset.")
+    }
     tres$pre_key=keys(tres, idcol="pre_id")
     tres$post_key=keys(tres, idcol='post_id')
     res[[n]]=tres
   }
-  if(isTRUE(bind.rows)) bind_rows2(res) else res
+  if(isTRUE(bind.rows)) {
+    res=bind_rows2(res)
+    # record the datasets we tried to find
+    attr(res, 'datasets')=names(ids)
+    res
+  } else res
 }
 
 

README.Rmd

@@ -11,6 +11,9 @@ knitr::opts_chunk$set(
   fig.path = "man/figures/README-",
   out.width = "100%"
 )
+# use released flywire cell types even if you have access to in progress
+# information
+options(fafbseg.use_static_celltypes = TRUE)
 ```
 
 
@@ -62,15 +65,22 @@ anticipated late 2024.
 
 ## Installation
 
-You can install the development version of **coconatfly** like so:
+We recommend installing the latest version of **coconatfly** from github like so:
 
-``` r
+```{r eval=FALSE}
 install.packages('natmanager')
 natmanager::install(pkgs = 'coconatfly')
 ```
 
-Some of the datasets exposed by **coconatfly** require authentication for access or
-are still being annotated in private pre-release. Please consult individual 
+but you may also choose a specific 
+[released version](https://github.com/natverse/coconatfly/releases) 
+for enhanced reproducibility:
+```{r eval=FALSE}
+natmanager::install(pkgs = 'coconatfly@v0.2.2')
+```
+
+Some of the datasets exposed by **coconatfly** require authentication for access
+or are still being annotated in private pre-release. Please consult individual 
 package dependencies for authentication details and do not be surprised if you
 do not have access to all datasets at this time.
 
@@ -96,7 +106,7 @@ library(dplyr)
 
 Two important functions are `cf_ids()` which allows you to specify a set of
 neurons from one or more datasets and `cf_meta()` which fetches information
-about the cell type. 
+about each neuron including its cell type.
 For example let's fetch information about DA1 projection neurons:
 
 ```{r}
@@ -191,7 +201,7 @@ cell type of two cells / hemisphere (i.e. they should not have been split into
 two cell types in the hemibrain).
 
 You can also see that cells from one hemibrain hemisphere often cluster slightly
-oddly (e.g. 387687146) - this is likely due to truncation of the axons or dendrites of these cells
+oddly (e.g. `387687146`) - this is likely due to truncation of the axons or dendrites of these cells
 or a paucity of partners from the left hand side of the hemibrain. 
 
 ## Going further
@@ -208,7 +218,7 @@ In addition in order to justify continued development of natverse tools in gener
 coconatfly in particular, we would appreciate two citations for
 
 * For the natverse: [Bates et al eLife 2020](https://doi.org/10.7554/eLife.53350)
-* For coconatfly: [Schlegel et al bioRxiv 2023](https://doi.org/10.1101/2023.06.27.546055)
+* For coconatfly: [Schlegel et al Nature 2024](https://doi.org/10.1038/s41586-024-07686-5)
 
 Should you make significant use of natverse packages in your paper
 (e.g. multiple panels or >1 figure), we would also strongly

README.md

@@ -60,13 +60,22 @@ anticipated late 2024.
 
 ## Installation
 
-You can install the development version of **coconatfly** like so:
+We recommend installing the latest version of **coconatfly** from github
+like so:
 
 ``` r
 install.packages('natmanager')
 natmanager::install(pkgs = 'coconatfly')
 ```
 
+but you may also choose a specific [released
+version](https://github.com/natverse/coconatfly/releases) for enhanced
+reproducibility:
+
+``` r
+natmanager::install(pkgs = 'coconatfly@v0.2.2')
+```
+
 Some of the datasets exposed by **coconatfly** require authentication
 for access or are still being annotated in private pre-release. Please
 consult individual package dependencies for authentication details and
@@ -97,8 +106,8 @@ library(dplyr)
 
 Two important functions are `cf_ids()` which allows you to specify a set
 of neurons from one or more datasets and `cf_meta()` which fetches
-information about the cell type. For example let’s fetch information
-about DA1 projection neurons:
+information about each neuron including its cell type. For example let’s
+fetch information about DA1 projection neurons:
 
 ``` r
 cf_meta(cf_ids('DA1_lPN', datasets = 'hemibrain'))
@@ -110,47 +119,44 @@ cf_meta(cf_ids('DA1_lPN', datasets = 'hemibrain'))
 #> 5  722817260 701 2435     2435       5635 Traced Roughly traced 1104413432
 #> 6  754534424 646 2364     2364       5309 Traced Roughly traced 1265805547
 #> 7  754538881 623 2320     2320       4867 Traced Roughly traced 1217284590
-#>   cropped  instance    type lineage notes  soma side class group   dataset
-#> 1   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA>  TRUE    R  <NA>  <NA> hemibrain
-#> 2   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA>  TRUE    R  <NA>  <NA> hemibrain
-#> 3   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA>  TRUE    R  <NA>  <NA> hemibrain
-#> 4   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA> FALSE    R  <NA>  <NA> hemibrain
-#> 5   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA> FALSE    R  <NA>  <NA> hemibrain
-#> 6   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA>  TRUE    R  <NA>  <NA> hemibrain
-#> 7   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA>  TRUE    R  <NA>  <NA> hemibrain
-#>             key
-#> 1 hb:1734350788
-#> 2 hb:1734350908
-#> 3 hb:1765040289
-#> 4 hb:5813039315
-#> 5  hb:722817260
-#> 6  hb:754534424
-#> 7  hb:754538881
+#>   cropped  instance    type lineage notes  soma side class subclass subsubclass
+#> 1   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA>  TRUE    R  <NA>     <NA>        <NA>
+#> 2   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA>  TRUE    R  <NA>     <NA>        <NA>
+#> 3   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA>  TRUE    R  <NA>     <NA>        <NA>
+#> 4   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA> FALSE    R  <NA>     <NA>        <NA>
+#> 5   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA> FALSE    R  <NA>     <NA>        <NA>
+#> 6   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA>  TRUE    R  <NA>     <NA>        <NA>
+#> 7   FALSE DA1_lPN_R DA1_lPN   AVM02  <NA>  TRUE    R  <NA>     <NA>        <NA>
+#>   group   dataset           key
+#> 1  <NA> hemibrain hb:1734350788
+#> 2  <NA> hemibrain hb:1734350908
+#> 3  <NA> hemibrain hb:1765040289
+#> 4  <NA> hemibrain hb:5813039315
+#> 5  <NA> hemibrain  hb:722817260
+#> 6  <NA> hemibrain  hb:754534424
+#> 7  <NA> hemibrain  hb:754538881
 ```
 
 We can also do that for multiple brain datasets
 
 ``` r
 da1meta <- cf_meta(cf_ids('DA1_lPN', datasets = c('hemibrain', 'flywire')))
-#> Updating 6641 ids
-#> flywire_rootid_cached: Looking up 6641 missing keys
-#> Updating 5480 ids
-#> flywire_rootid_cached: Looking up 5480 missing keys
+#> Loading required namespace: git2r
 head(da1meta)
-#>                   id side   class    type      lineage group  instance dataset
-#> 1 720575940604407468    R central DA1_lPN ALl1_ventral  <NA> DA1_lPN_R flywire
-#> 2 720575940623543881    R central DA1_lPN ALl1_ventral  <NA> DA1_lPN_R flywire
-#> 3 720575940637469254    R central DA1_lPN ALl1_ventral  <NA> DA1_lPN_R flywire
-#> 4 720575940614309535    L central DA1_lPN ALl1_ventral  <NA> DA1_lPN_L flywire
-#> 5 720575940617229632    R central DA1_lPN ALl1_ventral  <NA> DA1_lPN_R flywire
-#> 6 720575940619385765    L central DA1_lPN ALl1_ventral  <NA> DA1_lPN_L flywire
-#>                     key
-#> 1 fw:720575940604407468
-#> 2 fw:720575940623543881
-#> 3 fw:720575940637469254
-#> 4 fw:720575940614309535
-#> 5 fw:720575940617229632
-#> 6 fw:720575940619385765
+#>                   id side   class subclass   subsubclass    type      lineage
+#> 1 720575940604407468    R central     ALPN uniglomerular DA1_lPN ALl1_ventral
+#> 2 720575940623543881    R central     ALPN uniglomerular DA1_lPN ALl1_ventral
+#> 3 720575940637469254    R central     ALPN uniglomerular DA1_lPN ALl1_ventral
+#> 4 720575940614309535    L central     ALPN uniglomerular DA1_lPN ALl1_ventral
+#> 5 720575940617229632    R central     ALPN uniglomerular DA1_lPN ALl1_ventral
+#> 6 720575940619385765    L central     ALPN uniglomerular DA1_lPN ALl1_ventral
+#>   group  instance dataset                   key
+#> 1  <NA> DA1_lPN_R flywire fw:720575940604407468
+#> 2  <NA> DA1_lPN_R flywire fw:720575940623543881
+#> 3  <NA> DA1_lPN_R flywire fw:720575940637469254
+#> 4  <NA> DA1_lPN_L flywire fw:720575940614309535
+#> 5  <NA> DA1_lPN_R flywire fw:720575940617229632
+#> 6  <NA> DA1_lPN_L flywire fw:720575940619385765
 ```
 
 ``` r
@@ -275,10 +281,8 @@ seems to work very well for this purpose.
 
 ``` r
 cf_cosine_plot(cf_ids('/type:LAL0(08|09|10|42)', datasets = c("flywire", "hemibrain")))
-#> Updating 6641 ids
-#> Updating 5480 ids
-#> Matching types across datasets. Dropping 510/1052 output partner types with total weight 9007/24134
-#> Matching types across datasets. Dropping 793/1493 input partner types with total weight 11121/27588
+#> Matching types across datasets. Keeping 542/1052 output connections with total weight 15127/24134 (63%)
+#> Matching types across datasets. Keeping 700/1493 input connections with total weight 16467/27588 (60%)
 ```
 
 <img src="man/figures/README-lal-cosine-cluster-1.png" width="100%" />
@@ -293,7 +297,7 @@ constitute a single cell type of two cells / hemisphere (i.e. they
 should not have been split into two cell types in the hemibrain).
 
 You can also see that cells from one hemibrain hemisphere often cluster
-slightly oddly (e.g. 387687146) - this is likely due to truncation of
+slightly oddly (e.g. `387687146`) - this is likely due to truncation of
 the axons or dendrites of these cells or a paucity of partners from the
 left hand side of the hemibrain.
 
@@ -313,8 +317,8 @@ we would appreciate two citations for
 
 - For the natverse: [Bates et al eLife
   2020](https://doi.org/10.7554/eLife.53350)
-- For coconatfly: [Schlegel et al bioRxiv
-  2023](https://doi.org/10.1101/2023.06.27.546055)
+- For coconatfly: [Schlegel et al Nature
+  2024](https://doi.org/10.1038/s41586-024-07686-5)
 
 Should you make significant use of natverse packages in your paper
 (e.g. multiple panels or \>1 figure), we would also strongly appreciate

tests/testthat/test-cosine.R

@@ -33,3 +33,15 @@ test_that("cosine plots work", {
       expect_equal(cf_cosine_plot(lalids, min_datasets = 1, heatmap = F),
                    hc1)))
 })
+
+
+test_that("cosine plot with no partners", {
+  op <- options(fafbseg.use_static_celltypes=T)
+  on.exit(options(op))
+
+  suppressWarnings(
+    suppressMessages(
+      expect_warning(cf_cosine_plot(cf_ids(hemibrain = 'ORN_DA1', flywire = 'ORN_DA1'), threshold = 8),
+                     "no inputs.*flywire")
+    ))
+})