Light Microscopy Data

Light microscopy is the largest body of data on VFB by number of source studies. It answers a different question from EM: rather than what a neuron connects to, it tells you which genetic reagent will let you see or manipulate a cell type, and what else that reagent labels. VFB registers confocal image stacks from more than a hundred studies onto the standard templates, classifies what they label with Drosophila Anatomy Ontology terms, and links each driver back to its FlyBase record and stock centre.

Kinds of LM data

The distinction that matters when searching is between an image of a whole expression pattern — everything a driver labels — and an image of a single neuron or small clone picked out of that pattern.

Every one of these is registered to a template, so a single-neuron image from one study can be compared directly against an EM reconstruction from another, and a driver’s pattern can be scored for overlap against any region or neuron.

Data providers

VFB holds 85 light microscopy datasets. Each is grouped below by where its images were produced. For the large collections that is recorded in VFB, through the source cross-references carried by the images or the dataset’s own description; for the directly-deposited sets it is taken from the cited paper.

Records are the individuals VFB holds from a dataset — expression patterns, fragments, single neurons or painted domains. They count what VFB has loaded, not what a collection contains at source.

Only datasets whose images are loaded into VFB are listed. Dataset records still being loaded are left out until their imagery is in, so a dataset appearing here means there is something to look at. Every one has an attributed source.

Citations are recorded on each dataset’s own page on VFB rather than repeated here, so that they stay correct if a record is updated. Cite the original study, not VFB, when you use the images.

Expression annotations

Registration puts an image in the right place; annotation is what makes it queryable. VFB curators and pipelines record, for each driver, which anatomical structures it is expressed in, as ontology-classified assertions rather than free text. That is what lets a query for a cell type return the drivers that label it, and a query for a driver return everything it is known to hit.

These annotations come from two sources: curation of the published literature, and computed overlap between a registered expression pattern and the painted domains or neuron images in the same template space. They are annotations of what has been observed and recorded — a driver with no recorded expression in a region has not been shown to be absent there.

Finding LM data

• From a cell type. Open any neuron class and run the expression queries on its Term Info pane to get the drivers reported to label it.
• From a region. Open a neuropil and ask for the expression patterns that overlap it.
• From an image. Run NBLAST from a single-neuron image to find morphologically similar neurons, LM or EM, across every registered dataset.
• From a reagent name. Search the line directly — R81G11, VT061192, SS04495 — or the FlyBase identifier.
• Programmatically. VFB_connect exposes the same queries; see the APIs page.

Adult LM images are registered to JRC2018Unisex or JRC2018UnisexVNC unless the study predates them, in which case the older JFRC2 or Court2018 VNS template may be the only alignment available. Check the template shown on the image’s Term Info page before comparing coordinates across studies.

Data VFB does not hold

VFB indexes and registers LM images, and links out to the original collections for the raw, unregistered data. For adult FlyLight material that means flweb.janelia.org, gen1mcfo.janelia.org and splitgal4.janelia.org; VFB hosts the raw larval FlyLight images itself, at raw.larval.flylight.virtualflybrain.org. See FlyLight and the external resources page.