Giving scientists superpowers with MegaMap
Scientists are too often forced to tackle next-generation problems with last generation’s tools. We’re putting today’s best tech in the hands of our partners working to fight disease — starting with our world-leading screening platform run on software called MegaMap.
December 7, 2021 Colorblind-friendly mode

At Spring, we built a high-throughput screening platform that sheds light on hundreds of cell behaviors across thousands of conditions — all in one screen — to help our partners discover drugs for many different diseases.

Fed by an automated lab generating terabytes of data, our platform unlocks valuable biological signals from a single powerful assay by combining high-content imaging and proteomics.

Cell interactions
Cell death
Predicted age
T cells
Monocytes
Podosomes
Mitochondria
Choose a cell property
Cell interactions
Cell death
Predicted age
T cells
Monocytes
Podosomes
Mitochondria
Cell interactions
In images of untreated cells
A field of view captured by a high-content microscope, highlighting the quantification of the Cell-to-cell feature in a control condition.
A field of view captured by a high-content microscope, highlighting the quantification of the Cell-to-cell feature in a control condition.
Cell interactions
In images of treated cells
A field of view captured by a high-content microscope, highlighting the quantification of the Cell-to-cell feature in the Drug 093 condition.
A field of view captured by a high-content microscope, highlighting the quantification of the Cell-to-cell feature in the Drug 093 condition.
Hover to see cells in more detail.
50μm
Cell interactions
Drug 093
MegaMap says:
Treated samples showed a significant decrease in cell interactions compared to control.

At the center is software we’ve called MegaMap, a tool built to give scientists the superpowers that lay at the intersection of human intuition and machine learning.

Drug 097
Drug 616
Drug 633
Drug 818
Drug 829
Drug 545
Drug 722
Each row represents a treatment (a drug), and each column represents a feature (like metabolic activity or cytokine expression).
Cytokine measurements
A schematic view of Spring's MegaMap
An instantly filterable, sortable visualization of a drug screen, showing hundreds of feature measurements computed across thousands of potential therapies.
% T Cells
# T Cells
% Cytotoxic Lymphocytes
# Cytotoxic Lymphocytes
% Monocytes
# Monocytes
% Activated Monocytes
# Activated Monocytes
% Macrophage
# Macrophage
% Dendritic Cells
# Dendritic Cells
% Nuclei Damaged
# Nuclei Damaged
% Dying Macrophages
# Dying Macrophages
% Dying T Cells
# Dying T Cells
% Apoptotic Cells
# Apoptotic Cells
Fragmented Nucleus
Pyknotic Nucleus
Kidney-Shaped Nucleus
Area
Maximum Radius
Mean Radius
Perimeter
Compactness
Form Factor
Reticular
Fragmentary
Reticular
Fragmentary
Lymphocyte-Lymphocyte
Monocyte-Monocyte
Lymphocyte-Monocyte
IL-8
MCP1
IL-6
IL-17A
TNF alpha
IL-4
IL-10
IL-2
IL-1 beta
IFN-gamma
Activator 1 Signature
Activator 2 Signature
Control 1 Signature
Control 2 Signature
Control 3 Signature
Control 4 Signature
Control 5 Signature
Control 6 Signature
Control 7 Signature
Activator 1 Signature
Activator 2 Signature
Control 1 Signature
Control 2 Signature
Control 3 Signature
Control 4 Signature
Control 5 Signature
Control 6 Signature
Control 7 Signature
Podosome Formation
Podosome Deletion
Actin Contraction
M1 Monocyte Signature
M2 Monocyte Signature
Endocytic Uptake
Antigen Uptake
Activator 1 Score
Activator 2 Score
Control 1 Score
Control 2 Score
Control 3 Score
Control 4 Score
Control 8 Score
Control 9 Score
Activator 1 Score
Activator 2 Score
Control 1 Score
Control 2 Score
Control 3 Score
Control 4 Score
Control 8 Score
Control 9 Score

MegaMap is fueled by a suite of machine learning models that measures hundreds of high-content imaging and proteomics features. Combined, they open up new possibilities to build the most comprehensive cellular models of complex diseases in the world — with deeper understanding of drugs’ on-target actions, off-target effects, and safety signals.

With an explorable interface that unifies cellular function, morphology, metabolomics, proteomics, spatial interactions, and more, we built MegaMap to help scientists ask the hardest questions and get fast answers:

Drugs that suppress tumor growth
Drug
Aging Signature
IL-8
IL-6
IL-17A
TNF alpha
IL-1 beta
% T Cells
% Monocytes
% Macrophage
Young
(Control)
Drug 430
Drug 573
Drug 625
Drug 063
Drug 505
Drug 195
Drug 822
Cytokines
Cell composition
Matches
These drugs show increased similarity to the 'young' phenotype (in green) and decreases in inflammatory cytokines (in red).
Drugs that improve vaccine response
Drug
Activator 1 Signature
Activator 2 Signature
IL-8
IL-6
IL-17A
TNF alpha
IL-1 beta
Reticular
Fragmentary
Activator 1
Activator 2
Drug 864
Drug 430
Drug 414
Drug 633
Drug 616
Drug 101
Cytokines
Mitochondria
These drugs show increased similarity to one of two known immune system activators (left, in green), included as controls at top.
Drugs that inhibit inflammation
Drug
Inhibitor 1 Signature
IL-8
IL-6
IL-17A
TNF alpha
IL-1 beta
Area
Perimeter
Compactness
Form Factor
Inhibitor 1
Drug 483
Drug 103
Drug 362
Drug 444
Drug 209
Drug 565
Drug 803
Cytokines
Morphology
These drugs show consistent anti-inflammatory effects (reducing IL-1 beta, in red) and similarity to a known inhibitor of inflammation (in green).

MegaMap gives scientists the superpowers needed to resolve hard queries like those above, decode unprecedented amounts of high-content imaging and proteomics data, and assess the behavior of thousands of drugs using primary human samples.

They can map their hypotheses to specific, functional understanding, and learn from unbiased hypotheses surfaced by the machine across hundreds of cell properties. Get a feel for (a small sample of) its power below:

Drug
% T Cells
# T Cells
% Cytotoxic Lymphocytes
# Cytotoxic Lymphocytes
% Monocytes
# Monocytes
% Activated Monocytes
# Activated Monocytes
% Macrophage
# Macrophage
% Dendritic Cells
# Dendritic Cells
% Nuclei Damaged
# Nuclei Damaged
% Dying Macrophages
# Dying Macrophages
% Dying T Cells
# Dying T Cells
% Apoptotic Cells
# Apoptotic Cells
Fragmented Nucleus
Pyknotic Nucleus
Kidney-Shaped Nucleus
Area
Maximum Radius
Mean Radius
Perimeter
Compactness
Form Factor
Reticular
Fragmentary
Reticular
Fragmentary
Lymphocyte-Lymphocyte
Monocyte-Monocyte
Lymphocyte-Monocyte
IL-8
MCP1
IL-6
IL-17A
TNF alpha
IL-4
IL-10
IL-2
IL-1 beta
IFN-gamma
Aging Signature
Lymphocyte Signature
Monocyte Signature
Activator 1 Signature
Activator 2 Signature
Control 1 Signature
Control 2 Signature
Control 3 Signature
Control 4 Signature
Control 5 Signature
Control 6 Signature
Control 7 Signature
Activator 1 Signature
Activator 2 Signature
Control 1 Signature
Control 2 Signature
Control 3 Signature
Control 4 Signature
Control 5 Signature
Control 6 Signature
Control 7 Signature
Podosome Formation
Podosome Deletion
Actin Contraction
M1 Monocyte Signature
M2 Monocyte Signature
Endocytic Uptake
Antigen Uptake
Activator 1 Score
Activator 2 Score
Control 1 Score
Control 2 Score
Control 3 Score
Control 4 Score
Control 8 Score
Control 9 Score
Activator 1 Score
Activator 2 Score
Control 1 Score
Control 2 Score
Control 3 Score
Control 4 Score
Control 8 Score
Control 9 Score
Young
(Control)
Activator 1
(Control)
Activator 2
(Control)
Drug 394
Drug 310
Drug 523
Drug 497
Drug 860
Drug 720
Drug 939
Drug 103
Drug 133
Drug 991
Cell composition
Cell death
Nucleus
Morphology
Mitochondria
Cytokines
Aging signature
Monocyte signature
Lymphocyte signature
Cytoskeleton
T-cell similarity
Monocyte similarity
Hover, click, or tap to interact
The data used throughout this entire post is a de-identified subset of a high-throughput screen executed by Spring in 2021, modified via random alterations and additive noise to maintain confidentiality. Specifically, the aging control data has been altered and is included for illustrative purposes only.

Powered by a toolbelt full of computational gadgets

We regularly run high-throughput drug screens for our partners, capturing hundreds of millions of primary human immune cells [1] stimulated by thousands of conditions, resulting in hundreds of terabytes of imaging and proteomic data.

As drug developers, our goal is to build rapid understanding of these drugs’ actions, unexpected effects, and safety signals. MegaMap uses a packed toolbelt of novel computational tools to help us do so, all while learning from the terabytes of biological data generated by each experiment.

The first such tool is our automatic cell type classifier, which lets us run assays on heterogeneous sets of primary human immune cells:

Identifying activated monocytes...
Replay
A field of view captured by a high-content microscope, highlighting individual cells over time as they're identified by the system.
A field of view captured by a high-content microscope, highlighting individual cells over time as they're identified by the system.
Activated monocytes
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
T cells
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
Macrophages
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
An individual cell identified within the broader field of view.
Hover to see cells in more detail.

In addition to the cell classification tool above, we deploy deep learning models at single-cell resolution to build cell-by-cell phenotypic profiles covering an array of valuable phenotypes and functions, a few of which you see here:

Area
Circularity
Regularity
Intensity
Sort cells by...
Area
Circularity
Regularity
Intensity
Largest area
Smallest area

By applying these novel computational models to every cell in a population, we turn unstructured images into high-content, structured data, letting scientists easily query across biologically-relevant cellular functions.

Seeing things that humans might not

Cellular phenotypes vary in easily-interpretable dimensions such as size, shape, and color. But cells and their functions also manifest emergent patterns that are much harder for humans to even describe, let alone find on their own amongst hundreds of millions of cells.

We believe there are valuable biological signals in these hard-to-discover patterns. MegaMap and its toolbelt unlock these signals previously hidden in phenotypic screens — similar to the way DNA sequencing technology unlocked a trove of previously-hidden biological data.

MegaMap does so by arming scientists with our single-cell deep learning embeddings, [2] which they can use to search for connections they’d otherwise never see. Without specifying what they’re searching for, they can find connections between treatment conditions, spot relationships between donor populations, flag correlations between features, and more.

Get a sneak peek at their power below: see how these embeddings can find similar cells among a pool of thousands of different samples, without anybody defining what “similar cell” means.

Cell similarity search — sample 201278
A field of view captured by a high-content microscope. Hovering of any given cell in the field of view triggers a search (by similarity).
Hover over any cell to find similar cells.
50μm
Selected cell
The individual cell selected from the broader field of view.
Monocyte from sample 201278.

At right, nine matches for this phenotypic profile from Spring's ever-growing collection of high-content imaging data.Below, nine matches for this phenotypic profile from Spring's ever-growing collection of high-content imaging data.
Similar cells, different treatments
The individual cell selected from the broader field of view.
Selected
The first most similar cell to the selected cell.
Drug 652
The second most similar cell to the selected cell.
Drug 229
The third most similar cell to the selected cell.
Drug 194
The fourth most similar cell to the selected cell.
Drug 617
The fifth most similar cell to the selected cell.
Drug 521
The sixth most similar cell to the selected cell.
Drug 098
The seventh most similar cell to the selected cell.
Drug 325
The eighth most similar cell to the selected cell.
Drug 748
The ninth most similar cell to the selected cell.
Drug 390
Showing 9 most similar cells of 182,396 candidates.

These single-cell embeddings are the engine behind many columns you see in MegaMap. Using them, we’ve built deep learning models to capture a smorgasbord of cellular functions like immune cell interactions, subtypes of cell death, and cytokine expression for scientists’ perusal, relying on the machine to distinguish cellular phenotypes that humans may not otherwise understand.

We think of MegaMap as kind of like Star Trek’s tricorder — but for pointing at primary cell samples to decode complex biology, discover connections, and screen drugs.

Try asking a few questions and clicking around:

MegaMap, show me...
mTOR inhibitors
that
alter macrophage polarization
Drug
% T Cells
% Cytotoxic Lymphocytes
% Monocytes
% Activated Monocytes
% Macrophage
% Dendritic Cells
M1 Monocyte Signature
M2 Monocyte Signature
Drug 310
Drug 209
Drug 626
Drug 955
Drug 340
Drug 294
Drug 610
Drug 817
Cell composition
Polarization

Science led by humans

We build state-of-the-art tools so scientists can wield the superpowers that exist at the intersection of human scientific expertise and new computational technology.

From small details in MegaMap’s user interface to our strategic decisions, Spring’s work is guided by the goal of empowering scientists. In order for technology to radically improve drug discovery, it's not enough for a computational tool to be superhuman at interpreting massive amounts of high-dimensional data — it must unite this understanding with scientists’ intuition.

Much of MegaMap centers on this idea, including its cornerstone ability to combine human “phenotype curation” with machine learning results to uncover new targets and in a high-dimensional screen.

Target discovery via machine learning + human conviction
MegaMap’s screening results are sorted by “curated feature scores” (on the left) which are defined by our scientists for every screen. They develop their own ranking methodologies, fine-tuning the readouts that matter most for a given program.

These features that scientists curate in MegaMap are powered by the state-of-the-art machine learning tech described above. But the real output of our work — the advancement of targets and drugs — comes from a combination of technology with human scientists’ conviction.

This humble combination is key to accelerating the discovery and development of therapies for many diseases.

Giving scientists superpowers in the battle against disease

Today's most problematic diseases present notoriously complex biological problems — they are driven by complicated, interdependent biology that is often not (yet!) well understood.

Empowering the scientists working on these challenges is an enormous opportunity. Those working in this space deserve the very best tools. We’re here to put the world’s best technology in our partners', starting with MegaMap and its gadgets above.

Request a Demo

If you're part of a group fighting back against disease and are excited by such tools, request a demo or contact us. We'd love to arm you with the technology you deserve.

Footnotes
  1. We use this system to study multiple cell types from many organs (and always primary samples), but our interest in immune aging has us focused on peripheral blood mononuclear cells (PBMCs), a heterogeneous immune cell population containing T cells, B cells, natural killer cells, monocytes, and more. ↩︎
  2. An embedding can be thought of as a ‘fingerprint’ of a given cell, a vector of numbers that capture an opaque latent representation of the cell’s phenotypic characteristics. ↩︎

Data sources
The data used throughout this entire post is a de-identified subset of a high-throughput screen executed by Spring in 2021, modified via random alterations and additive noise to maintain confidentiality. Specifically, the aging control data has been altered and is included for illustrative purposes only. Similarly, all example images were chosen to maximize clarity and may not align with any purported labels.


Acknowledgements
Thanks to the following for feedback on drafts of this post: Matt Booty, Christian Elabd, Ben Komalo, Lauren Nicolaisen, Tim Sullivan, and Brandon White.

Appendix

High-content imaging

For those not familiar: in high-content immunofluorescence imaging, every field of view contains from zero to hundreds of cells (depending on magnification, cell type, etc). And for every field of view, we capture up to six fluorescent channels, each of which is meant to light up a different slice of cell biology. Get a feel for these different channels by playing with the sliders below.

Red
On
Green
On
Blue
On
By default, we assign one immunofluorescent stain to each of red, green, and blue.
Red
Off
Green
Off
Blue
On
Isolating to the Hoechst stain, assigned to the blue channel, we see each cell's nucleus.
Red
On
Green
Off
Blue
Off
Concanavalin A, in red, stains carbohydrates on cellular membranes.
Red
Off
Green
On
Blue
Off
Phalloidin, in green, stains the actin cytoskeleton.

When we visualize this kind of data as humans, it’s common for Hoechst — which stains the cell’s nucleus — to be visualized in blue, while Phalloidin — which stains the cell’s actin cytoskeleton — could be represented in green. These choices are more conventions than anything, though, as any channel can be lit up in any color.

AboutTeamTechnologyNews
PartnershipsSpring's tech is used by a range of partners across biotech, pharma, and academic research. We provide both strategic collaborations and software licensing to those looking for novel biological insights or tools to accelerate their science. Contact us.
© 2023 Spring Discovery. All rights reserved.