Team:Lambert GA/Software

• TEAM
  • TEAM MEMBERS
  • ATTRIBUTIONS
• PROJECT
  • DESCRIPTION
  • DESIGN
  • CONTRIBUTION
  • PROPOSED IMPLEMENTATION
  • ENGINEERING SUCCESS
  • PROOF OF CONCEPT
  • EXCELLENCE IN ANOTHER AREA
  • SUSTAINABLE DEVELOPMENT
  • PARTNERSHIP
• WETLAB
  • PARTS LIBRARY
  • PARTS DESCRIPTION
  • PART COLLECTION
  • PHO
  • NAR
  • PPB
  • EXPERIMENTS
  • NOTEBOOK
  • RESULTS
• DRYLAB
  • HARDWARE
  • MEASUREMENT
  • MODEL
• HUMAN PRACTICES
  • INTEGRATED HUMAN PRACTICES
  • EDUCATION
  • SCIENCE COMMUNICATION
  • COLLABORATIONS
  • SAFETY
• JUDGING FORM
• POSTER

SOFTWARE

PLATE-Q

OVERVIEW

Plate-Q is a frugal plate reader capable of quantifying green fluorescence protein (GFP) and Optical Density (OD) from a 96-well standard plate. A typical laboratory plate reader can cost up to $15,000, which is a major bottleneck for smaller-scale labs that are unable to afford such equipment. Plate-Q costs under $150 and can be adapted to scan for different types of fluorescent proteins at different wavelengths. Rather than using traditional optical sensors, such as photodiodes, found in laboratory microplate readers, Plate-Q takes advantage of a Raspberry Pi camera to capture images of a well plate and extract image features using computer vision and machine learning algorithms. A 440 nm wavelength excitation light is used to measure GFP fluorescence using a 510 nm filter for emission, and a 600 nm light source is used for OD without any emission filter. Plate-Q is completely open-source and users can customize the design to scan for other fluorescent proteins by replacing the light source and filter for different wavelengths. Users can retrain the Plate-Q algorithm to take measurements in an affordable, cost-effective manner.

PARTS

• Frosted Acrylic Plate:
• Raspberry Pi
• Raspberry Pi Camera with wide-angle lens + high sensitivity
• 440 nm LED Strip Lights:
• 600 nm LED Strip Lights
• Reflective material of light
• 510 nm emission filter

DESIGN

3D Plate Reader Case:

Plate-Q is broken down into 3 detachable parts: Top cover (Fig. __), a Light-diffusing chamber (Fig. __), and a bottom cap (Fig. __). The raspberry camera is housed on the top cover and has 4 positions of movement. The cover is printed with a matte black filament which helps prevent the reflection of the light into the camera. This part then slides onto the light-diffusing chamber. The 96-well plate is placed into a slot on top of the light-diffusing chamber. We use a frosted acrylic diffuser that is able to diffuse the light from the LED strips evenly across the well plate. The bottom cap, which houses the led light strips, will be under the diffuser in order to provide the GFP and OD600 wavelengths needed for quantification.

SOFTWARE PROCESSING

Plate-Q relies on a standardized Raspberry Pi camera for the quantification of fluorescence and optical density, which is done through a software pipeline that extracts the image features and maps those values to laboratory-grade plate reader values. The pipeline starts by taking pictures in triplicates at 6 different positions across the well plate. Each image is then converted into a black and white image to analyze the perceived brightness of the entire image. Each well is cropped to the region of interest, and the pixel values within this region are averaged to get the brightness value. Then, the values of the triplicates are averaged to eliminate the outliers and the final value is stored in a data frame (CSV file) that is passed through a machine-learning algorithm to output the laboratory-grade plate reader values. The wells are pin-pointed using an algorithm that averages the brightness value across the well on a scale of 0 being the darkest to 255 being the lightest. This is performed for each image in the triplicate, then each well’s values are averaged and stored in a data frame.

Image of Pipeline:

MACHINE LEARNING TRAINING

Plate-Q relies on a machine-learning algorithm to accurately output values that correspond to traditional plate reader values. To train this algorithm, 50 well plates were filled with 100 µL of fluorescein in each well. Images of the well plates were captured in triplicates and passed through the software pipeline shown above to get the brightness of each well in arbitrary units. All the brightness values from each trial were stored in a master data frame. These well plates were also scanned inside a traditional plate reader to get the corresponding laboratory-grade fluorescence values. These fluorescence readings were also appended into the master data frame. The same process was repeated with an OD calibration fluid, using a Lead Chloride (PbCl2) precipitate solution, which was pipetted into all wells, scanned in Plate-Q and a traditional plate reader, and recorded in a master data frame. This data frame was then fed into the software processing pipeline in order to model a relationship between the brightness values obtained by Plate-Q and the emission values recorded by the laboratory plate reader.

RESULTS