Imaging 2D Gels: Resolution, Bit Depth and Dynamic Range
The quality of a 2D gel analysis is set before the analysis starts, at the scanner. An image captured at the wrong resolution, saturated in the bright spots, or saved as an 8-bit JPEG has thrown away information that no software can recover. This guide covers what to set and why: imaging mode, resolution, bit depth, dynamic range, saturation and file format, with a checklist to run through before every scan. It applies equally to conventional stained gels, 2D-DIGE fluorescence images and 2D Western blots, and to any analysis package, including our own SameSpots and SpotMap 2D.
Pre-scan checklist for 2D gel imaging
Run through this before every scan. Each item is explained in the sections that follow.
- Image in grayscale, not color.
- Capture at 16 bits per pixel. Never save quantitative images at 8 bits.
- Use the full grayscale range for each dye or stain without saturating any spot.
- Choose a resolution at which the smallest spots you care about are at least 5 pixels across.
- Keep orientation identical across every image in the experiment.
- Position every gel the same way and scan the same region of interest.
- Save the imaging parameters as a profile and reuse it for every gel in the experiment.
- In a multiplexed (DIGE) experiment, set parameters separately for each dye and keep them fixed.
- Do no post-processing beyond cropping, mirroring and 90-degree rotation. Your analysis software can do these.
- Save in a lossless, calibrated format: the scanner’s native GEL or IMG/INF format where available, otherwise 16-bit TIFF. Never JPEG.
Choosing an imaging mode
2D gel imagers work in one of three modes, and the right one depends on how the gel was stained.
Transmission. Light passes through the gel and is detected on the far side. This is the standard mode for visible stains such as Coomassie and silver, on flatbed densitometers and CCD-based gel documentation systems. It gives the most linear response and is the least prone to saturation, and it is what we recommend where the stain allows it.
Reflection. Light passes through the gel, reflects off a backing, and passes back through the gel before detection. It is used for opaque supports such as membranes and for some office-type scanners. The light path through the gel is doubled, so it saturates more readily.
Fluorescence. The dye is excited at one wavelength and its emission is detected at a longer one. This is the mode for 2D-DIGE (Cy2, Cy3, Cy5), for fluorescent stains such as SYPRO Ruby, and for fluorescent Western blots. Fluorescence imagers are either laser scanners with photomultiplier tube detectors or CCD camera systems with filtered LED illumination. Laser scanners typically offer higher resolution and dynamic range; CCD systems are faster and more versatile.
Whichever instrument you use, image with the instrument’s own acquisition software and save the raw output. Image-editing applications are not designed for quantitative data and can change pixel values in ways you cannot see.
Resolution: how many pixels a spot needs
Spot detection software finds a spot by its outline and quantifies it by summing the pixels inside. A spot represented by two or three pixels has no outline to find. The rule is that the smallest spot you intend to analyze should be at least 5 pixels in diameter, and more is better up to the point where file size becomes a problem.
Scanner resolution is quoted in dots per inch (dpi) or in microns per pixel. The two are related: 25,400 divided by the dpi gives the pixel size in microns. What that means for spots:
| Resolution | Pixel size | A 1 mm spot is | A 0.5 mm spot is |
|---|---|---|---|
| 100 dpi | 254 µm | about 4 pixels | 2 pixels: too few |
| 150 dpi | 169 µm | about 6 pixels | 3 pixels: too few |
| 200 dpi | 127 µm | about 8 pixels | about 4 pixels |
| 300 dpi | 85 µm | about 12 pixels | about 6 pixels |
| 600 dpi | 42 µm | about 24 pixels | about 12 pixels |
For a large-format gel where the spots of interest are around a millimeter across, 200 to 300 dpi is sufficient. For mini gels, where spots are smaller, 300 dpi is the minimum and 600 dpi is often needed. Bear in mind that every doubling of resolution quadruples the file size, so a 600 dpi scan of a large gel produces a very large image that will slow down analysis without improving it.
Spot detection on a 20cm 2D gel image, captured at a resolution of (a) 100dpi and (b) 300dpi. The lower resolution of the 100 dpi image is apparent by the degree of pixilation. In this image, there are fewer pixels present to represent the spots (approximately 2 orders of magnitude less pixels in the 100 dpi image compared to the 300 dpi image).
Bit depth: 16-bit vs 8-bit
Bit depth is the number of gray levels each pixel can take:
| Bit depth | Gray levels |
|---|---|
| 8-bit | 256 |
| 10-bit | 1,024 |
| 12-bit | 4,096 |
| 16-bit | 65,536 |
To the eye an 8-bit and a 16-bit image of the same gel look identical, because a screen cannot show more than a few hundred gray levels. To the software they are not identical at all. A faint spot and its background may differ by a handful of levels out of 256 in an 8-bit image and by hundreds of levels out of 65,536 in a 16-bit image. The 16-bit image can quantify that spot; the 8-bit image cannot separate it from noise.
Capture at 16 bits per pixel for any image you intend to quantify. Most scientific imagers output 16-bit natively; the common mistake is exporting or converting to 8-bit afterward, often by saving as JPEG or by opening the file in a general-purpose editor. SameSpots checks bit depth when images are imported and flags 8-bit images before analysis begins.
Dynamic range
Bit depth is the number of gray levels available. Dynamic range is how many of them you actually use. An image whose darkest pixel is 0 and whose brightest is 1,000 on a 16-bit scale is using 1.5% of the available range, and the faint spots are crushed into a few levels near the bottom.
The aim is to spread the gel’s intensities across as much of the range as possible without touching the top. Adjust exposure time on a CCD system, or photomultiplier voltage on a laser scanner, until the brightest spot is close to but below the maximum. Scan only the region of interest, because empty gel and the gel edge add nothing and can dominate the histogram.
Do not try to fix dynamic range after the scan. Stretching the histogram in an editor spreads the same few levels over a wider range; it does not create information, and it changes the relationship between pixel value and protein amount that quantification depends on.
Saturation
A pixel is saturated when the signal exceeds the maximum value the detector can record. It is clipped to the maximum, and everything above that is lost. In a 2D gel image, saturation flattens the top of the most abundant spots into plateaus.
Why saturation is a problem
Spot detection works on the shape of a spot: a peak with an outline. A saturated spot has no peak, only a flat top, so the software may draw an incorrect outline, split it, or merge it with neighbors. Quantification is worse: the volume of a saturated spot is underestimated by an unknown amount, and if the saturated spots are the abundant ones, the normalization of the whole gel is affected.
Recommendation
Set exposure so that the brightest spot on the gel sits just below the maximum. If a few very abundant spots must be saturated to see the faint ones, take two scans at different exposures, and note which spots are unreliable in the high-exposure image. SameSpots image quality control flags saturated regions on import so they can be reviewed before analysis.
File formats
Save in the scanner’s own calibrated format where one exists. Formats such as GEL (Cytiva and Amersham laser scanners) and IMG/INF (Fujifilm) store intensity with a calibration that analysis software can read, and SameSpots and SpotMap 2D import them directly. Where there is no native format, save as 16-bit uncompressed TIFF.
Never save a quantitative image as JPEG. JPEG compression is lossy, it discards exactly the small intensity differences that faint spots depend on, and it is limited to 8 bits per channel. A JPEG of a gel is a picture of the data, not the data.
Consistency across an experiment
Everything above matters more across a set of gels than for any one gel, because the analysis compares gels to each other. Two gels scanned at different resolutions, exposures or orientations differ for reasons that have nothing to do with the biology, and the software has to work harder to remove that difference.
Save the acquisition settings as a named profile in the scanner software and apply it to every gel in the experiment. Scan gels in the same orientation, positioned the same way on the platen, cropped to the same region of interest. Scan replicates in the same session where possible. When importing into SameSpots, image quality control checks resolution, bit depth, file size and orientation across the dataset and highlights any image that does not match.
Frequently asked questions
Q: What does each spot represent in a 2D gel?
A: Each spot is a protein, or a single isoform of a protein, that has been separated by isoelectric point horizontally and by molecular weight vertically. The position of the spot gives the protein’s approximate pI and mass; the intensity of the spot is proportional to the amount of that protein in the sample. Some spots contain more than one protein, which is why identification by mass spectrometry is still needed to be certain.
Q: Should I image 2D gels at 8-bit or 16-bit?
A: 16-bit. An 8-bit image has 256 gray levels and cannot separate faint spots from background reliably. A 16-bit image has 65,536 levels. The two look the same on screen; they are not the same to the analysis software.
Q: What resolution should I scan a 2D gel at?
A: Enough that the smallest spot you want to analyze is at least 5 pixels across. For large-format gels with millimeter-sized spots that is 200 to 300 dpi; for mini gels 300 to 600 dpi.
Q: Can I use JPEG for gel images?
A: Not for quantification. JPEG is lossy and 8-bit. Use the scanner’s native format or 16-bit TIFF.
Q: What is saturation in a gel image?
A: Pixels whose signal exceeds the detector’s maximum are clipped to that maximum. Saturated spots have flat tops, cannot be quantified accurately and can disturb spot detection.
Q: Can I adjust brightness and contrast before analysis?
A: No. Adjusting brightness, contrast or levels changes pixel values and breaks the relationship between intensity and protein amount. Cropping, mirroring and 90-degree rotation are the only safe edits, and the analysis software can do them for you.
Q: Which scanner do I need for 2D gels?
A: For visible stains, a transmission-mode flatbed densitometer or a CCD gel documentation system. For 2D-DIGE and fluorescent stains, a fluorescence imager: a laser scanner with photomultiplier detection or a CCD system with filtered illumination. SameSpots is vendor-neutral and accepts images from any of them.
Check your images before you waste time analyzing them
SameSpots runs image quality control on every image you import: bit depth, saturation, resolution and file size consistency, and orientation. Problems are flagged before you spend time on analysis, not after.