Skip to content

Latest commit

 

History

History
280 lines (210 loc) · 12.4 KB

NEWS.md

File metadata and controls

280 lines (210 loc) · 12.4 KB

fafbseg 0.10.0

The key change is to react to a breaking change in the URL to access spine services.

  • Feature flytable_append_rows by @jefferis in natverse#132
  • Update Buhmann ref date by @emilkind in natverse#138
  • Feature/misc flywire updates by @jefferis in natverse#139
  • Add flywire_nuclei to pkgdown by @jefferis in natverse#140
  • More flytable enhancements / fixes by @jefferis in natverse#141
  • Better handling of NAs in when handling body ids by @jefferis in natverse#142
  • Fix/spine rename by @jefferis in natverse#146
  • spine.janelia.org -> services.itanna.io by @perlman in natverse#144
  • changed top.nt and top.p to top_nt and top_p by @alexanderbates in natverse#143

New Contributors

Full Changelog: https://github.com/natverse/fafbseg/compare/v0.9.4...v0.10.0

fafbseg 0.9.4

  • add support for the Seung/Allen python CAVE including flywire_cave_client() flywire_cave_query() (#122)
  • add flywire_updateids() (#127)
  • also more robustness to handling of NAs in flywire_rootid()
  • first implementation of flytable_* functions including flytable_query() flytable_list_rows() flytable_login() (#126)
  • note about SSL certificate problem with some flywire/fanc queries bug (#125)
  • Add ids type info to flywire_partners()

https://github.com/natverse/fafbseg/issues?q=closed%3A2021-09-20..2021-10-29+

fafbseg 0.9.3 (pre-release version)

This is a pre-release without full release notes, principally to support a new dependant package, fancr.

  • new flywire_scene() function for simple specification of flywire neuroglancer scenes.
  • add a number of features to allow programmatic manipulation of neuroglancer scenes. This includes ngl_segments<-() to replace the segments in a scene +.ngscene() and -.ngscene() to add or remove segments from a scene and as.character.ngscene() to convert a scene object to a URL. See examples in ngl_segments() and ngl_decode_scene() for further details (#55, #58).
  • update flywire_set_token() to support new cave and zetta tokens (for FANC dataset) (#119)
  • expose chunkedgraph_token() function (#119)
  • Further updates to token handling to handle FANC/FlyWire (#121)
  • Private function check_cloudvolume_reticulate now supports a minimum version
  • Handle 1 node neurons (#120)
  • Basic support for the Python caveclient package. This enables fetching a range of annotation data from the CAVE (Connectome Annotation Versioning Engine) API including position of nuclei or user annotations. One notable improvement is more rapid fetching of synaptic partner information. This is implemented by the new method="cave" option for flywire_partner_summary() (#122).

https://github.com/natverse/fafbseg/issues?q=closed%3A2020-12-27..2021-09-20+

fafbseg 0.9

The version change to 0.9 reflects some small but potentially breaking changes in user visible behaviour. In particular ngl_segments() now returns as character by default (to protect FlyWire ids, which cannot be expressed as 8 byte floating point values).

  • Add flywire_latestid() function to map any (past) root id to the current root id. Specifically it it identifies the neuron inheriting the largest number of super voxels from the object specified by the input root id.
  • Support for Google ffn1 20200412 segmentation (#14, #46)
  • Make ngl_segments() return ids as character vector, rather than numeric, by default. Also exclude hidden ids by default (#50, #51).
  • Allow ngl_segments() (and flywire_*) fns to accept URLs with surface meshes. Previously these would cause an error (#52)
  • Fix ngl_encode_url() so that all tested flywire scenes can be parsed in R and then converted back to valid URLs that can be opened in flywire.ai (#53)
  • Add print.ngscene method to summarise neuroglancer scenes in the console (#54)
  • flywire_cloudvolume() enables improved low-level support for flywire+cloudvolume (by exposing a reusable cloudvolume object) (#48)

full list of closed issues on GitHub.

fafbseg 0.8.2

  • Much faster flywire_xyz2id() supervoxel id mapping using spine service (#44)
  • Add new flywire_leaves() function to find supervoxels in a neuron (#43)
  • Fixing typos and adding doi for TEASAR (#41 by @mmc46)
  • Fixing 2 things in overlay mesh+skeleton example (#40 by @mmc46)
  • Fix flywire_xyz2id eg in doc 'The natverse and flywire meshes' returns an error (#39)
  • Output of dr_fafbseg() points to wrong token helper function (#38 by @mmc46)

You can see the full list of closed issues on GitHub.

fafbseg 0.8.1

  • New function skeletor() to skeletonise meshes especially flywire neurons based on the Python package of the same name (#35).
  • Large speed up for flywire_rootid() (#37)
  • Fix flywire_xyz2id() so that it can actually find supervoxels (as well as root ids) (#37)
  • function dr_fafbseg() to give a status update on your installation (#36)
  • read_cloudvolume_meshes() can now accept flywire URLs (which will be expanded if necessary to define the segments to download)

You can see the full list of closed issues on GitHub.

fafbseg 0.8.0

  • Support for some basic flywire API calls #31, #34
  • Better integrate flywire api features
  • Fix/flywire coordinate errors
  • Fix error in fafb2flywire() during xform_brain #30, #32, #33
  • Switch to transform-service API on spine
  • Add support for FANC3-FANC4 transforms #28
  • switch back to wire3d #27
  • Add fafb2flywire() i.e. the inverse of the original FlyWire->FAFB transformation #26
  • Support for reading flywire meshes without cloudvolume #25
  • Flywire API streamlining (@SridharJagannathan) #22
  • Teach read_cloudvolume_meshes() to work for flywire URLs #20
  • ngl_segments() needs to be able to expand flywire URLs #16

You can see the full list of closed issues on GitHub.

fafbseg 0.7.0

  • basic support for reading meshes via CloudVolume e.g. to fetch neurons from flywire.ai. See read_cloudvolume_meshes() for details including the required Python setup.
  • very basic support for skeletonising neurons meshed (e.g. flywire neurons) via meshparty_skeletonize()
  • add flywire2fafb() to support for transforming neurons FlyWire->FAFB (and with some loss of accuracy in the reverse) (#11)
  • includes a fix for NaN values in return (#12,#13)
  • much faster FlyWire->FAFB transformation, up to millions of points per minute (#24)
  • moved repo to https://github.com/natverse/fafbseg (#19)

fafbseg 0.6.5

  • default chunksize for brainmaps_xyz2id() reduce to 200 to reflect API changes.
  • new functions choose_segmentation() and with_segmentation() to choose default auto-segmentation
  • New Shiny app (see README and https://jefferislab.shinyapps.io/CATMAID-Neuroglancer-Converter/)
  • give brainmaps_fetch() a generic cache option and the ability to clear the cache with brainmaps_clear_cache() (#8)
  • give brainmaps_fetch() a retry option to help with sporadic timeouts (#9)
  • give catmaid2ngl() a chunksize option that can be used to reduce timeout issues.
  • simplify package .onLoad and retire fafbseg.divisor option (now calculated automatically rather than being a user option)
  • simplify internal brainmaps_voxdims function using cache mechanism

fafbseg 0.6.4

  • fix bug revealed by latest public version of zip package
  • fix missing import of dplyr::n() in find_topn()

fafbseg 0.6.3

  • Give find_topn() and read_top() ability to return segments in increasing (rather than decreasing) size order.

fafbseg 0.6.2

  • fix ngl_encode_url() when only one segment in scene (#5)
  • fix handling or neuroglancer URLs with two segment sources (#4)

fafbseg 0.6.1

  • New catmaid2ngl() function and methods, a high level approach to converting URLs, neurons etc to representations based on Neuroglancer / brainmaps data.

fafbseg 0.6.0

  • default remote volume for brainmaps_xyz() has been updated to brainmaps://772153499790:fafb_v14:fafb_v14_16nm_v00c_split3xfill2_flatreseg2_skeletons32
  • this volume is also used for read.neurons.brainmaps()

fafbseg 0.5.5

  • brainmaps_skeleton() and read.neurons.brainmaps() now use a brainmaps URI to specify the remote skeleton source.
  • new package option fafbseg.skeletonuri to specify default remote skeleton source
  • give brainmaps_xyz2id() a chunksize argument to handle more query points than the remote API will accept in a single call.

fafbseg 0.5.4

  • add read.neuron.brainmaps() and read.neurons.brainmaps() to read skeletons into nat neuron objects over the web.
  • this depends on the lower level brainmaps_skeleton() function.
  • teach find_merge_groups() to return segment ids

fafbseg 0.5.3

  • compare_ng_neuron(): add pointsize and sample_dots args
  • doc / example tweaks

fafbseg 0.5.2

  • Support for reading 3D meshes directly from brainmaps API via read_brainmaps_meshes()

fafbseg 0.5.1

  • export brainmaps_xyz2id() (#2)

fafbseg 0.5.0

  • Add brainmaps_auth(), brainmaps_fetch() to talk to brainmaps API
  • Add brainmaps_xyz2id() to convert arbitrary XYZ locations to the corresponding segment id.
  • Teach find_merged_segments() to return the merge groups

fafbseg 0.4.1

  • additional functions for reading/processing merge (agglomeration) information including read_mergeinfo(), make_merge_graph(), merge_graph_components(). These were used to generate the fafbsegdata package and could be useful in their own right.

fafbseg 0.4.0

  • Add find_merged_segments() to find all the raw segments that are candidate merges from the agglomeration runs. This functionality depends on the new fafbsegdata package, which should be installed automatically as a suggested dependency.
  • Switch default zip file divisor to 1E6 (and actually check the contents of a zip file to accommodate a new default in Peter Li's 2018-10-02 skeleton release (fafb14_v00c_split3xfill2x_skeleton32nm512_nnconn75_thresh1000_sparse250)

fafbseg 0.3.5

  • fix bug in read_segments() datafrac argument logic - was only reading largest fragment
  • datafrac only applies to files > minfilesize
  • add internal skelsforsegments() function with progress and use it in read_segments2()
  • Turn on zip_list memoisation again but with a 5 min cache timeout - can speed up read_segments2 considerably when there are multiple segments inside the same zip file.

fafbseg 0.3.4

  • turn off zip_list memoisation to save memory (#1)

fafbseg 0.3.3

  • teach read_segments2() to read only the top n percent of skeletons by file size
  • additional option to read coordinates only
  • fix handling of zip files with >65535 files (see r-lib/zip#11) by making use of new ziplist64 package when available

fafbseg 0.3.2

  • export ngl_decode_scene() function to parse neuroglancer URLs or JSON scene specifications (in turn allowing these to be opened in CATMAID)
  • fail better when skeleton zip files are missing
  • fix bug in reading multiple bit64 ids

fafbseg 0.3.1

  • export open_fafb_ngl()
  • teach open_fafb_ngl to produce coordinate string to paste into Neuroglancer
  • add ngl_segments() to extract segment ids from diverse scene specifications such as URLs or JSON fragments copied from Neuroglancer web page

fafbseg 0.3.0

  • read_segments2() (in memory zip extraction) recommended
  • add find_topn() to find biggest segments in zip files
  • teach read_topn() to cope with multiple zip files
  • fix sample URL to point to correct segmentation (fafb_v14_16nm_v00c_split3xfill2) that is compatible with the skeletons we have received.

fafbseg 0.2.0

  • add support for reading neurons from zipped skeleton files
  • support for mapping ids to files and vice versa
  • simplify setup for neuroglancer URLs
  • improved docs

fafbseg 0.1.0

  • first version