**3. Methods**

### *3.1. Public Database and Literature Searches*

The search strategy employed in this review is summarized in Figure 2. Initially, the MGI database [34] was queried to identify mutant protein-coding genes associated with a mammalian disease phenotype indicating a loss of PR cells. MGI is a curated database that includes expert annotation based on full-text searching of 148 selected journals, which is limited compared to literature databases, such as PubMed. Typically, only the first paper describing a new allele is fully curated for phenotype data in the database due to resource constraints. Although our analysis yielded 159 mutant genes that were associated with PR cell loss, a number of mutant genes known to cause this phenotype were absent or not annotated, possibly due to these aforementioned limitations.

**Figure 2.** Flow chart depicting the progression of the search strategy utilized in this review. The number of records or genes identified from them is indicated at each stage of the process. The dashed line indicates that some genes were identified from records that remain to be systematically screened.

To expand the search, we used NCBI databases, including PubMed [35] and Gene. We refined a PubMed query by searching with keyword phrases to generate article lists, using the Gene option of Find Related Data to yield mouse genes linked to the articles, and then assessing whether these genes were on our MGI list. The goal was to develop a broad query that included as many genes from the MGI list as possible but also included additional hits. The most successful was: (ONL OR "outer nuclear layer" OR retina\* OR PR OR rod OR rods OR cone\*) AND (degener\* OR loss OR thin\* OR thick\*) AND (mouse OR mice OR murine), which captured >97% of the MGI list. Restricting this query to entries posted to PubMed on or before October 15, 2019, yielded 9535 articles. To review these articles efficiently we tried two approaches, the first generating a spreadsheet containing hyperlinks in which each linked gene symbol was combined with the Boolean query, and the second using mouse gene identification numbers corresponding to the linked genes and applying an Entrez script that accessed the Gene and PubMed databases to find all articles satisfying the Boolean query for each linked gene.
