Eye in a Disk (EiaD)

EiaD is the sqlite database at the core of eyeIntegration.nei.nih.gov

For 2023 (versions >= 2.0), we have updated the "backend" of EiaD in several significant ways:

1. Simplify the salmon-based quantification to better enable integration of our dataset with outside resources
2. Added many new samples and studies
3. More granular metadata schema that has five major categories:

• Tissue (e.g. Retina)
• Sub_Tissue (e.g. Macula)
• Source (e.g. tissue or iPSC)
• Age (e.g. fetal or adult)
• Perturbation (e.g. None or AMD)

4. Added ML-based sex labels
5. Built a recount3 based quantification pipeline (http://github.com/davemcg/Snakerail) to enable base pair level coverage information
6. Used ML based approach to identify sample outliers for QC
7. Summarized cell type level gene tables imported from our plae.nei.nih.gov resource

Workflow

1. Snakerail (http://github.com/davemcg/Snakerail) wraps the pump (output) and unify (RSE) steps in monorail (https://github.com/langmead-lab/monorail-external)
2. Snakefile runs a salmon-based quant to generate gene and transcript level counts
3. There are four "hand" steps that generate the 2023 EiaD datasets that are run in this order:

• scripts/pull_scEiaD.R and scripts/build_eiad_2023_plae.R
• scripts/metamoRph_label.R and scripts/identify_outlier_samples.Rmd
• scripts/pca_workup_data_prep.R
• scripts/build_eiad_2023_bulk.R