Sharing Analyzed Datasets

Every plate of yeast that was grown during each screen had four control lines grown at n=4, and many plates also included blank wells. Additionally, we recorded the Optical Density (OD) of each sample before it was processed. All of this data can be used to normalize the data from each sample and plate. However, different analysts may have different views on the most appropriate way to integrate these controls into the analysis and determine what they consider to be significant changes. We have designed Piims to accommodate multiple processed versions of the same raw data, all of which will be presented to a user searching the database, along with a short description of the type of analysis. Users uploading their won analysis will be asked for a detailed description of their methods. The database will take them through a flexible process that defines the data fields in their analysis and allows for aggregation of data using many different parameters. Users will upload as many of the following files as appropriate, and the uploaded dataset will be reviewed by the Ionomicshub team to ensure it meets quality and accuracy standards (our goal is not to keep datasets out, we will work with users to help them meet the standards).

• 1. Normalized data
• 2. Z-score or other statistical value to measure statistically relevant change
• 3. % difference or some other value to measure magnitude of the difference
• 4. P-values to evaluate significance of a given change
• 5. All data for the type in 1, 2 & 3 will have values for individual samples and/or summary values for a line (mean, median or mode) and a measure of variation (st dev, IQR).
• 6. Collect metadata that describes the type of analysis done on the dataset

Users will be able to use Piims data display, graphing and downloading tools on any dataset in the system.