Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 73 of the 74 references it cites were found in a registry or an index, and 1 are not checked yet; 41 of the 56 sources its claims cite were checked for support.

FlyWire: online community for whole-brain connectomics

Open access

Published version of a preprint

Abstract

Due to advances in automated image acquisition and analysis, whole-brain connectomes with 100,000 or more neurons are on the horizon. Proofreading of whole-brain automated reconstructions will require many person-years of effort, due to the huge volumes of data involved. Here we present FlyWire, an online community for proofreading neural circuits in a Drosophila melanogaster brain and explain how its computational and social structures are organized to scale up to whole-brain connectomics. Browser-based three-dimensional interactive segmentation by collaborative editing of a spatially chunked supervoxel graph makes it possible to distribute proofreading to individuals located virtually anywhere in the world. Information in the edit history is programmatically accessible for a variety of uses such as estimating proofreading accuracy or building incentive systems. An open community accelerates proofreading by recruiting more participants and accelerates scientific discovery by requiring information sharing. We demonstrate how FlyWire enables circuit analysis by reconstructing and analyzing the connectome of mechanosensory neurons.

Verified 2026-11-09 · 1 reference not checked yet

Overview

CoChat Summary

TL;DR

FlyWire is an open online community and computational platform for proofreading a whole-brain Drosophila connectome, built on a ChunkedGraph data structure and browser-based 3D editing. It demonstrates scalable proofreading and circuit discovery by reconstructing mechanosensory pathways, estimating roughly 19 minutes of human effort per neuron.

Key findings

  • The ChunkedGraph made edit operations over an order of magnitude faster than a naive supervoxel graph; server response times during beta were median 47 ms for supervoxel-to-root (95th percentile 111 ms, n=12,096) and median 48 ms for root-to-supervoxels per displayed chunk (95th percentile 465 ms, n=3,080,494). Merge edits had median 940 ms (95th percentile 1,841 ms, n=4,612); split edits had median 1,818 ms (95th percentile 7,137 ms, n=2,497).
  • Proofreading 183 neurons with 13 proofreaders over three rounds yielded median F1 scores of 0.99 for volumetric completeness and 0.99 for synapse assignment after one round; a faster regimen took a median of 13 minutes per cell, with mean proofreading time 19.1 minutes. A whole-brain backbone connectome for ~100,000 cells was estimated at 16 person-years.
  • In 178 mechanosensory neurons across seven cell classes, the authors identified 59 left and 58 right AMMC-B1 neurons, versus prior work identifying only 10 per hemisphere; only 14 of 59 left and 14 of 58 right AMMC-B1 neurons innervated WED-VLPs, and AMMC-B1 neurons divided into at least five connectivity-based subtypes.
  • Synapse omission from orphan twigs was quantified at 612 randomly picked synaptic locations: 40.6% of postsynaptic and 78.2% of presynaptic twigs were attached to backbones, leading to an estimate that 44.6% of synapses had both pre- and postsynaptic segments attached after proofreading; major connections (>9 synapses) were maintained at 99.7% with at least one synapse, and most minor connections with at least 3 synapses at 83%.

Methods

  • Design: an open online proofreading community using a ChunkedGraph supervoxel graph with octree hierarchy, Google BigTable storage, and a Neuroglancer-based browser frontend; the FAFB EM dataset was realigned from v14 to v14.1 using self-supervised deep learning displacement fields.
  • Data and sample: adult Drosophila melanogaster FAFB EM volume and automated segmentation; 183 neurons proofread by 13 proofreaders in three rounds; 178 mechanosensory neurons analyzed; 612 synaptic locations assessed for twig attachment; 999 synapses expert-evaluated; NBLAST comparison against 16,129 FlyCircuit cells.
  • Analysis: server response timing for graph operations; F1 scores for volumetric completeness and synapse assignment; NBLAST matching; connectivity diagrams and subtype classification by synapse thresholds; max-flow min-cut for split operations.

Limitations

  • Stated: proofreading largely avoided orphan twigs, so synapses involving orphan twigs are missing; false positive synapses remain among reported cell-type-to-cell-type connections; some left-hemisphere AMMC-B1 neurons lacked part of their arbor, possibly a developmental deformity; whole-brain proofreading time assumes ~100,000 neurons and 2,000 work hours per person-year; beta server timings were affected by load and are an underestimate.
  • Stated: the open community relies on entry tests and subsequent spot checks, and recruitment initially came from Drosophila labs, with plans to expand to non-scientists later.
  • Evident from the text: the connectome is a draft proofread mainly at the backbone level, with accuracy depending on the user’s desired proofreading level; the mechanosensory analysis is limited to one fly brain and selected cell classes, and some connectivity calls depend on manual exclusion of questionable synapses.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/s41592-021-01330-0

Cites

73 of 74 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Assessing segmentation quality using known neurons. ( a-d ) Comparison of light-level stains of known giant fiber neurons ( a ) and a mushroom body APL neuron ( b , red) to FlyWire's AI-predicted segmentation of these cells ( c,d ). ( e,f ) The same neurons shown following proofreading. ( g-n ) Examples of other known cell types before and after proofreading (top and bottom of each pair, respectively): central complex neurons ( g,h ), olfactory projection neurons ( i,j ), gustatory receptor neurons ( k,l ) and a lobula plate tangential cell ( m,n ). All views frontal except APL and central complex neurons: dorso-frontal view. Scale bars: (c, d, e, f, i, j) 30 μm, (g, h, k, l) 15 μm, (m, n) 20 μm
Figure 1 Assessing segmentation quality using known neurons. ( a-d ) Comparison of light-level stains of known giant fiber neurons ( a ) and a mushroom body APL neuron ( b , red) to FlyWire's AI-predicted segmentation of these cells ( c,d ). ( e,f ) The same neurons shown following proofreading. ( g-n ) Examples of other known cell types before and after proofreading (top and bottom of each pair, respectively): central complex neurons ( g,h ), olfactory projection neurons ( i,j ), gustatory receptor neurons ( k,l ) and a lobula plate tangential cell ( m,n ). All views frontal except APL and central complex neurons: dorso-frontal view. Scale bars: (c, d, e, f, i, j) 30 μm, (g, h, k, l) 15 μm, (m, n) 20 μmFrom the PDF
Fig. 2 Proofreading the supervoxel graph. (a) Automated segmentation overlaid on the EM data. Each color represents an individual putative neuron. (b) Each component in (a) is made up of many supervoxels shown in different colors. (c) This panel corresponds to the framed square in (a) and the full panel in (b) where we highlighted the supervoxels belonging to a particular neuron and overlaid a cartoon of the supervoxel graph. (d) Touching supervoxels (circles) may be connected through edges in the graph indicating that they belong to the same connected component (solid lines). Merge operations add edges between supervoxels resulting in new neuronal components (orange). (e) Split operations remove edges resulting in new neuronal components (green, blue). (f) Example neuron after proofreading (black). Green, blue and red components were removed during proofreading (see Ext. Data Fig. 4 for the added pieces). While edit operations have global effects, the edits to the supervoxel graph themselves are performed at a local level. (g) For splits, users place points (red and blue dots) either in 2D or 3D (center panel) that are linked to the underlying supervoxels (left panel). The proofreading backend then automatically determines which edges need to be removed and performs the split (right panel). (h) For the operation shown in (g) the backend performs max-flow min-cut on the local supervoxel graph to determine the optimal cut that separates the user-defined input locations (green and blue framed circles). The thickness of the edges symbolizes the edge weight (cartoon). Scale bars (a,b,c): 1μm (f): 10 μm
Fig. 2 Proofreading the supervoxel graph. (a) Automated segmentation overlaid on the EM data. Each color represents an individual putative neuron. (b) Each component in (a) is made up of many supervoxels shown in different colors. (c) This panel corresponds to the framed square in (a) and the full panel in (b) where we highlighted the supervoxels belonging to a particular neuron and overlaid a cartoon of the supervoxel graph. (d) Touching supervoxels (circles) may be connected through edges in the graph indicating that they belong to the same connected component (solid lines). Merge operations add edges between supervoxels resulting in new neuronal components (orange). (e) Split operations remove edges resulting in new neuronal components (green, blue). (f) Example neuron after proofreading (black). Green, blue and red components were removed during proofreading (see Ext. Data Fig. 4 for the added pieces). While edit operations have global effects, the edits to the supervoxel graph themselves are performed at a local level. (g) For splits, users place points (red and blue dots) either in 2D or 3D (center panel) that are linked to the underlying supervoxels (left panel). The proofreading backend then automatically determines which edges need to be removed and performs the split (right panel). (h) For the operation shown in (g) the backend performs max-flow min-cut on the local supervoxel graph to determine the optimal cut that separates the user-defined input locations (green and blue framed circles). The thickness of the edges symbolizes the edge weight (cartoon). Scale bars (a,b,c): 1μm (f): 10 μmFrom the PDF
Fig. 3 The ChunkedGraph approach for proofreading supervoxel graphs. (a) One-dimensional representation of the supervoxels graph. In the simplest approach (naive), one would store the connected component information (neuronal component) in a dedicated parent node. (b) We propose an alternative data structure to enable proofreading of very large supervoxel graphs. We store the connected component information in an octree structure where each abstract node (black and colored nodes in levels > 1) represents the connected component in the spatially underlying graph (dashed lines represent chunk boundaries, see Ext. Fig. 3). Nodes on the highest layer represent entire neuronal components. (c) Edits to the ChunkedGraph (here, a split; indicated by the red arrow and removed red edge) do not require the loading of the entire supervoxel graph to recompute the neuronal connected components. We can reuse the connected component information stored in the abstract nodes that did not see a change to their underlying graph. (d) Chunks double in size along each dimension when going to the higher layer. (e) Server response times for the remapping of the connected components from root to supervoxel (N=3,080,494) and supervoxel to root (N=12,096) (f) as well as splits (N=2,497) and merges (N=4,612) for real user interactions in the beta-phase of FlyWire. (g) Compared to the naive approach (Fig. 3a), our approach reduces the number of supervoxels that need to be loaded for a split (global vs local) (h) The subgraph can be read faster and (i) speeds up max-flow min-cut calculations (Fig. 2g,h). The red lines in (g, h, i) are mean and the shaded area standard deviation of bins along x-axis (10 bins); N=15,233.
Fig. 3 The ChunkedGraph approach for proofreading supervoxel graphs. (a) One-dimensional representation of the supervoxels graph. In the simplest approach (naive), one would store the connected component information (neuronal component) in a dedicated parent node. (b) We propose an alternative data structure to enable proofreading of very large supervoxel graphs. We store the connected component information in an octree structure where each abstract node (black and colored nodes in levels > 1) represents the connected component in the spatially underlying graph (dashed lines represent chunk boundaries, see Ext. Fig. 3). Nodes on the highest layer represent entire neuronal components. (c) Edits to the ChunkedGraph (here, a split; indicated by the red arrow and removed red edge) do not require the loading of the entire supervoxel graph to recompute the neuronal connected components. We can reuse the connected component information stored in the abstract nodes that did not see a change to their underlying graph. (d) Chunks double in size along each dimension when going to the higher layer. (e) Server response times for the remapping of the connected components from root to supervoxel (N=3,080,494) and supervoxel to root (N=12,096) (f) as well as splits (N=2,497) and merges (N=4,612) for real user interactions in the beta-phase of FlyWire. (g) Compared to the naive approach (Fig. 3a), our approach reduces the number of supervoxels that need to be loaded for a split (global vs local) (h) The subgraph can be read faster and (i) speeds up max-flow min-cut calculations (Fig. 2g,h). The red lines in (g, h, i) are mean and the shaded area standard deviation of bins along x-axis (10 bins); N=15,233.From the PDF
Figure 4 Attaching automatically detected synapses to neurons. (a) Each edit (black dot) is linked to a user and timestamp enabling the retrieval of the edit history and credit assignment post-hoc. Neurons in the fly brain can be roughly partitioned into sections known as 'twigs' and 'backbone'. (b) We classified pre- and (c) post-synaptic segments based on their morphology and whether they are attached to a bigger component that will be attached during a conservative procedure. (d) shows examples of these assessments. (e) shows a proofread WV-WV PN (top) and AMMC-B1 (bottom) with automatically detected synapses attached (blue: presynaptic, red: postsynaptic). Scale bar: (d): 1 μm; (e) 25 μm
Figure 4 Attaching automatically detected synapses to neurons. (a) Each edit (black dot) is linked to a user and timestamp enabling the retrieval of the edit history and credit assignment post-hoc. Neurons in the fly brain can be roughly partitioned into sections known as 'twigs' and 'backbone'. (b) We classified pre- and (c) post-synaptic segments based on their morphology and whether they are attached to a bigger component that will be attached during a conservative procedure. (d) shows examples of these assessments. (e) shows a proofread WV-WV PN (top) and AMMC-B1 (bottom) with automatically detected synapses attached (blue: presynaptic, red: postsynaptic). Scale bar: (d): 1 μm; (e) 25 μmFrom the PDF
Figure 5 Proofreading in FlyWire. (a) Number of edits per neuron (N=183) and proofreading round. (b) Same as (a) but restricted to large edits (> 1μm 3 ). Large edits are usually edits on 'backbones' whereas small edits tend to affect 'twigs'. (c) Comparing the F1-Scores (0-1, higher is better; with respect to proofreading results after three rounds) between different proofreading rounds according to volumetric completeness and (d) assigned synapses. 'Auto' refers to reconstructions without proofreading.
Figure 5 Proofreading in FlyWire. (a) Number of edits per neuron (N=183) and proofreading round. (b) Same as (a) but restricted to large edits (> 1μm 3 ). Large edits are usually edits on 'backbones' whereas small edits tend to affect 'twigs'. (c) Comparing the F1-Scores (0-1, higher is better; with respect to proofreading results after three rounds) between different proofreading rounds according to volumetric completeness and (d) assigned synapses. 'Auto' refers to reconstructions without proofreading.From the PDF
Figure 6 Auditory connectivity extracted with FlyWire. (a) We included 180 neurons from both hemispheres into our analysis, here colored by their cell type (black: giant fiber neuron, turquoise: AMMC-A1, orange: AMMC-A2, dark blue: AMMC-B1, light blue: AMMC-B2, red: WED-VLP PN, green: WV-WV PN). (b) Connectivity diagram between all 180 neurons ordered by cell type. Gray through lines divide cells from different hemispheres (left/top: left hemisphere, right/bottom: right hemisphere) and colored bars separate putative sub-celltype (see Extended Data Fig. 6 for matrices with different colormap cutoffs). (c) Analyzing the synaptic inputs of the WED-VLP neurons onto the suggested groups shows differential input from AMMC-A2 subtypes. (d) The two WED-VLP groups differ in their morphology (see arrow heads). (e) We grouped AMMC-B1 neurons according to their outputs on to other cell types in the AMMC. (f) Abridged summarized connectivity diagram showing the selective targeting of WED-VLP PNs (WV) by AMMC-B1s (B1) and the cross-pathway innervation by AMMC-B1-4s. (g) The connectivity based groupings of AMMC-B1 neurons correspond to a stration of axonal arbors in both hemispheres (arrow heads). (h) One group of AMMC-B1 neurons targets AMMC-A1 neurons in both hemispheres (red: AMMC-A1, turquoise: AMMC-B1-4, gray: all AMMC-A1s). We automatically detected 217 synapses from this specific AMMC-B1 neuron onto all AMMC-A1 neurons (balls), 87 of which on the shown AMMC-A1 (green balls). (i) Commissural neurons vary in their connectivity with other cell types. (j) The suggested subgrouping correlates with collocations of their cell bodies as well as different arborizations. Scale bars: 25 μm
Figure 6 Auditory connectivity extracted with FlyWire. (a) We included 180 neurons from both hemispheres into our analysis, here colored by their cell type (black: giant fiber neuron, turquoise: AMMC-A1, orange: AMMC-A2, dark blue: AMMC-B1, light blue: AMMC-B2, red: WED-VLP PN, green: WV-WV PN). (b) Connectivity diagram between all 180 neurons ordered by cell type. Gray through lines divide cells from different hemispheres (left/top: left hemisphere, right/bottom: right hemisphere) and colored bars separate putative sub-celltype (see Extended Data Fig. 6 for matrices with different colormap cutoffs). (c) Analyzing the synaptic inputs of the WED-VLP neurons onto the suggested groups shows differential input from AMMC-A2 subtypes. (d) The two WED-VLP groups differ in their morphology (see arrow heads). (e) We grouped AMMC-B1 neurons according to their outputs on to other cell types in the AMMC. (f) Abridged summarized connectivity diagram showing the selective targeting of WED-VLP PNs (WV) by AMMC-B1s (B1) and the cross-pathway innervation by AMMC-B1-4s. (g) The connectivity based groupings of AMMC-B1 neurons correspond to a stration of axonal arbors in both hemispheres (arrow heads). (h) One group of AMMC-B1 neurons targets AMMC-A1 neurons in both hemispheres (red: AMMC-A1, turquoise: AMMC-B1-4, gray: all AMMC-A1s). We automatically detected 217 synapses from this specific AMMC-B1 neuron onto all AMMC-A1 neurons (balls), 87 of which on the shown AMMC-A1 (green balls). (i) Commissural neurons vary in their connectivity with other cell types. (j) The suggested subgrouping correlates with collocations of their cell bodies as well as different arborizations. Scale bars: 25 μmFrom the PDF
Image 7
From the PDF
Extended Data Figure 2. Quality of EM image alignment. We calculated the chunked pearson correlation (CPC) between two neighboring sections in the original alignment (v14) and our re-aligned data (v14.1). (a) The relative change of CPC between the original and our re-aligned data per section (b) is almost always positive (dashed red line is at 0). (c, d, e) show examples from (a) where (c) the CPC improved through a better alignment around an artifact, (d) the CPC is almost identical and (e) the CPC overall improved due to a stretch of poorly aligned sections in the original data that were resolved in v14.1.
Extended Data Figure 2. Quality of EM image alignment. We calculated the chunked pearson correlation (CPC) between two neighboring sections in the original alignment (v14) and our re-aligned data (v14.1). (a) The relative change of CPC between the original and our re-aligned data per section (b) is almost always positive (dashed red line is at 0). (c, d, e) show examples from (a) where (c) the CPC improved through a better alignment around an artifact, (d) the CPC is almost identical and (e) the CPC overall improved due to a stretch of poorly aligned sections in the original data that were resolved in v14.1.From the PDF
Extended Data Figure 3. Chunking the dataset. (a) Automated segmentation overlayed on the EM data. Each different color represents an individual putative neuron. (b) The underlying supervoxel data is chunked (white dotted lines) such that each supervoxel is fully contained in one chunk. (c) A close up view of the box in (b). (d) We apply the same chunking scheme to the meshes, requiring only minimal mesh recomputations after edits. (e) The irregular shape (not a box) of the fly brain leads to a high diversity of the number of supervoxels in each chunk (median: 25661). (f) The median supervoxel contains 792 voxels. All very small supervoxels (< 200 voxels) are the result of chunking.
Extended Data Figure 3. Chunking the dataset. (a) Automated segmentation overlayed on the EM data. Each different color represents an individual putative neuron. (b) The underlying supervoxel data is chunked (white dotted lines) such that each supervoxel is fully contained in one chunk. (c) A close up view of the box in (b). (d) We apply the same chunking scheme to the meshes, requiring only minimal mesh recomputations after edits. (e) The irregular shape (not a box) of the fly brain leads to a high diversity of the number of supervoxels in each chunk (median: 25661). (f) The median supervoxel contains 792 voxels. All very small supervoxels (< 200 voxels) are the result of chunking.From the PDF
Extended Data Figure 4. Proofreading with the ChunkedGraph. (a, b) Analogous to the split operation in Fig. 3a, b a merge operation does not require the loading and writing to and from the entire connected component since we can reuse the connected component information stored in the abstract nodes that did not see a change to their underlying graph. (c) During proofreading, all shown segments were added together. This is the same neuron for which removed segments were shown in Fig 2. Scale bar (c): 10 μm
Extended Data Figure 4. Proofreading with the ChunkedGraph. (a, b) Analogous to the split operation in Fig. 3a, b a merge operation does not require the loading and writing to and from the entire connected component since we can reuse the connected component information stored in the abstract nodes that did not see a change to their underlying graph. (c) During proofreading, all shown segments were added together. This is the same neuron for which removed segments were shown in Fig 2. Scale bar (c): 10 μmFrom the PDF
Extended Data Figure 5. The FlyWire proofreading platform. ( a ) The most common view in FlyWire displays the EM image in grayscale overlaid with segmentation in color (left panel), a 3D view of selected cell segments (middle), and menus with multiple proofreading tools (right). ( b ) Annotation tools include points, which can be used for a variety of purposes such as marking particular cells or synapses.
Extended Data Figure 5. The FlyWire proofreading platform. ( a ) The most common view in FlyWire displays the EM image in grayscale overlaid with segmentation in color (left panel), a 3D view of selected cell segments (middle), and menus with multiple proofreading tools (right). ( b ) Annotation tools include points, which can be used for a variety of purposes such as marking particular cells or synapses.From the PDF
Image 12
From the PDF
Extended Data Figure 7. Connectivity diagrams. (a) Diagram from Figure 6b reordered by subtype (b) Same diagram as in Figure 6b with different colormap threshold.
Extended Data Figure 7. Connectivity diagrams. (a) Diagram from Figure 6b reordered by subtype (b) Same diagram as in Figure 6b with different colormap threshold.From the PDF

Verification

Claims the paper makes

20 Valid · 9 Not valid · 8 Not checked · 98 Own results and interpretations

50 more claims · 4 with their cited sources checked

Create a free account to see all 135 claims and every source verdict.

See all claims — free account
  • Probably not a substantial claim. 36
  • It may state two separate claims as one. 54
  • The quote only partly states this claim. 1

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

10 more claims about this paper

Create a free account to see every claim other papers make about this one.

See all claims — free account

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

2 of 2

Citations

Cites

Show all 66 references

Research works printed but not linked: 1

  1. Mu, S. et al. 3D reconstruction of cell nuclei in a full Drosophila brain. Preprint at bioRxiv https://doi.org/10.1101/2020.04.17.047167 (2021). · DOI · Paper page

Cited by

See all 297

Metadata

Authors
Dorkenwald, S., McKellar, C. E., Macrina, T., Kemnitz, N., Lee, K., Lu, R., Wu, J., Popovych, S., Mitchell, E., Nehoran, B., Jia, Z., Bae, J. A., Mu, S., Ih, D., Castro, M., Ogedengbe, O., Halageri, A., Kuehner, K., Sterling, A. R., Ashwood, Z., Zung, J., Brittain, D., Collman, F., Schneider-Mizell, C., Jordan, C., Silversmith, W., Baker, C., Deutsch, D., Encarnacion-Rivera, L., Kumar, S., Burke, A., Bland, D., Gager, J., Hebditch, J., Koolman, S., Moore, M., Morejohn, S., Silverman, B., Willie, K., Willie, R., Yu, S.-C., Murthy, M., & Seung, H. S.
Affiliations
  • Princeton University (US)
  • Allen Institute for Brain Science (US)
Corresponding author
  • Murthy, Mala
  • Seung, H. Sebastian
Venue
Nature Methods
Published
2021-12-23
Volume
19
Issue
1
Pages
119-128
ISSN
  • 1548-7091 (ISSN-L)
  • 1548-7105
OpenAlex
W4206755867
Cited by
289
References
66
Citations by year
2026: 45 · 2025: 59 · 2024: 74 · 2023: 50 · 2022: 36 · 2021: 17
FWCI
20.06
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Methods (publishedVersion)
  • bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by-nd)
  • PubMed (publishedVersion)
  • PubMed Central (acceptedVersion)
Funders
  • ODNI | Intelligence Advanced Research Projects Activity (D16PC00004, D16PC0005, D16PC00003, D16PC00005)
  • G. Harold and Leila Y. Mathers Foundation
  • Amazon Web Services
  • Google
  • U.S. Department of Health & Human Services | NIH | National Institute of Mental Health (RF1 MH117815-01, RF1MH123400-01, PRIME NIH MH #5-U01MH117072 (MIT sub), 5U01MH114824-04 (CSHL subaward))
  • Howard Hughes Medical Institute (Faculty Scholar Award)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Biophysics › Cell Image Analysis Techniques
  • Life Sciences › Neuroscience › Cellular and Molecular Neuroscience › Neurobiology and Insect Physiology Research
  • Social Sciences › Education › Undergraduate Neuroscience Education and Research
Keywords
Drosophila melanogaster brain, online collaborative editing, edit history, incentive mechanisms, mechanosensory neurons, neural circuit reconstruction, FlyWire
MeSH terms
Data Visualization, Animals, Brain, Computer Graphics, Drosophila melanogaster, Neurons, Software, Imaging, Three-Dimensional, Connectome

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-11-09, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-09 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: FlyWire: online community for whole-brain connectomics. https://app.cochat.ai/papers/10.1038/s41592-021-01330-0
The paper
Dorkenwald, S., McKellar, C. E., Macrina, T., Kemnitz, N., Lee, K., Lu, R., Wu, J., Popovych, S., Mitchell, E., Nehoran, B., Jia, Z., Bae, J. A., Mu, S., Ih, D., Castro, M., Ogedengbe, O., Halageri, A., Kuehner, K., Sterling, A. R., . . . Seung, H. S. (2022). FlyWire: online community for whole-brain connectomics. Nature Methods. https://doi.org/10.1038/s41592-021-01330-0

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.