The workflow for using Pourfecto is:

1. Define source labware containing reagent stocks.
2. Define target labware.
3. Select available instrument configurations.
4. Run the Pourfecto planning and scheduling algorithm to create a solved Pourcast
5. Compile and Inspect the resulting Pourcast.

Defining Source and Labware

The easiest way to define the source and target labware inputs is to use Pourfecto's DataFrame interface.

using Pourfecto, CSV, DataFrames

# load properly formatted source DataFrames 
source_value_df = CSV.read("<source_value_file>.csv",DataFrame)
source_unit_df = CSV.read("<source_unit_file>.csv",DataFrame)
# generate source labware 
sources = df_to_labware(source_value_df,source_unit_df)
# load target DataFrames
target_value_df = CSV.read("<target_value_file>.csv")
target_unit_df = CSV.read("<target_unit_file>.csv")
# generate target labware 
targets = df_to_labware(target_value_df,target_unit_df)

Selecting Available Configurations

Pourfecto defines liquid handler instances as Configuration objects. Configurations combine an instrument's pipetting Head with a Deck that can hold the source and target labware. Users can define custom configurations, but Pourfecto provides an assortment of default configurations in the configurations dictionary.

In this example we will select two defualt Configurations:

1. an eight channel pipette (oriented in the vertical direction)
2. a Hamilton Nimbus (configured with slots for 50 mL conical tubes and a single SLAS plate slot, a common configuration in the Jensen Lab)

 
available_configs = [configurations["eight_channel_vertical"],configurations["nimbus"]] 

Running the Pourfecto algorithm

Pourfecto's main function, pourfecto, is a flexible method for running the Pourfecto algorithm and compiling Pourcasts. Given an output directory, pourfecto creates an optimal liquid handling plan and schedule, saves it as a Pourcast, and compiles the Pourcast into executable instrument files. When run in this mode, the pourfecto method automatically checks the solution quality before compiling, and throws errors if the solution falls outside a pre-specified tolerance.

pourcast = pourfecto("<output_directory>", sources, targets, available_configs) 

Pourcasts can also be generated without providing an output directory. In this case, the pourcast must be manually compiled.

pourcast = pourfecto(sources,targets,available_configs) 
compile("<output_directory>",pourcast) 

Inspecting Solutions

pourfecto creates the following file structure when provided with an <output_directory>

<output_directory>/
├── pourcast.json
├── target_plate_images/
│   ├── <plate name 1>.png
│   └── ... 
├── <Configuration 1>/
│   ├── <protocol 1>/  
│   │   ├── loading_instructions.png 
│   │   ├── loading_table.csv
│   │   └── instrument files ... 
│   ├── <protocol2>/  ... 
│   └── ... 
├── <Configuration 2>/
│   ├── <protocol 1>/  
│   │   ├── loading_instructions.png 
│   │   ├── loading_table.csv
│   │   └── instrument files ... 
│   └── ...
└── ...

• pourcast.json: The Pourcast is automatically saved in a .json format
• target_plate_images/: A heatmap of each target plate is generated showing the planned final volume of each well, which is helpful for verifying solutions visually.
• <Configuration>/: Instrument files are written into Configuration-specific folders, where each subfolder within is an executable protocol with a randomly generated name. Protocol subfolders each contain instrument loading instructions in table and image formats.