Design mixed meals automatically for maximizing the information content of their glycan compositions.
- Gurobi (version ≥ 9.1)
- R (version ≥ 4.0.2)
- R Packages gurobi, MaxPro, ggplot2, readxl, stringr, argparse, infotheo, reshape2
Install Gurobi (version ≥ 9.1). The Gurobi academic licence is free. Then run the following command (ATTENTION: use the location of the Gurobi R package on your system which varies depending on your installation):
export GUROBI_R_PATH=$HOME/local/gurobi912/linux64/R/gurobi_9.1-2_R_4.0.2.tar.gz
Ensure that you have conda installed (see miniconda), then run:
conda create -n omms -c conda-forge -c ameenetemady r-omms
conda activate omms
Run: setup_OMMs.R --gurobi-R-package $GUROBI_R_PATH
Design optimal mixed meals (OMMs) to maximize the information content of glycan profiles. In other words, in terms of the glycan content, we want the meals to be most different from each other while also being most different from the individual foods. To run, use the following command (change the argument values as needed):
generate_OMMs.R --glycanDB glycanDB.csv --moistureDB moistureDB.xlsx --num-meals N [--max-meal-ingredients M] --output generated_OMMs.csv
--glycanDB filename.csv: The glycan content profiles of individual foods when dried. The format is described in ./data/glycanDB_format.txt.--moistureDB filename.xlsx: The moisture percentages of individual foods before drying. The format is described in ./data/moistureDB_format.txt.--max-meal-ingredients M(optional): The maximum number of ingredients in each designed mixed meal.--compositional(optional): Transform glycan content of each food to compositional values. If not set, glyans will be transformed using minimax normalization for each glycan independently.--num-meals N: The number of mixed meals to design.--output filename.csv: The designed OMMs where each mixed meal is defined by the individual food proportions that it includes. The format is described in ./data/MMs_format.txt.
Select a set of OMMs from candidate mixed meals, given a set of approved mixed meals. To run, use the following command (change the argument values as needed):
select_OMMs.R --glycanDB glycanDB.csv --moistureDB moistureDB.xlsx [--approved-meals approved_meals.csv] --candidate-meals candidate_meals.csv --num-meals N --output selected_OMMs.csv
--glycanDB filename.csv: The glycan content profiles of individual foods when dried. The format is described in ./data/glycanDB_format.txt.--moistureDB filename.xlsx: The moisture percentages of individual foods before drying. The format is described in ./data/moistureDB_format.txt.--num-meals N: The number of mixed meals to select.--approved-meals filename.csv(optional): The set of approved meals. The format is described in ./data/MMs_format.txt.--compositional(optional): Transform glycan content of each food to compositional values. If not set, glyans will be transformed using minimax normalization for each glycan independently.--candidate-meals filename.csv: The set of candidate meals to select from. The format is described in ./data/MMs_format.txt.--output filename.csv: The selected OMMs where each mixed meal is defined by the individual food proportions that it includes. The format is described in ./data/MMs_format.txt.
Visualize the expected glycan content of OMMs. Figure1 will include a stacked barchart visualizing the glycan content of each OMM. Figure2 will show the information content of the OMMs' expected glycan profiles when measured sequentially. To run, use the following command (change the argument values as needed):
visualize_OMMs.R --glycanDB glycanDB.csv --moistureDB moistureDB.xlsx --OMMs OMMs.csv --output-dir ./resutls
--glycanDB filename.csv: The glycan content profiles of individual foods when dried. The format is described in ./data/glycanDB_format.txt--moistureDB filename.xlsx: The moisture percentages of individual foods before drying. The format is described in ./data/moistureDB_format.txt--OMMs filename.csv: The OMMs to visualize. The format is described in ./data/MMs_format.txt.--output-dir dir_path: The output directory for saving the generated figures.
See examples.sh. Following a successful installation, you can run ./test/examples.sh.
For any questions contact Ameen Eetemadi (eetemadi@ucdavis.edu).
See the LICENSE file for license rights and limitations (MIT License).
This work was supported by the United States Department of Agriculture (USDA)/NSF AI Institute for Next Generation Food Systems (AIFS), USDA award number 2020-67021-32855.