Figure 2

Dependence of the relative enrichment in literature-inferred gene roles as a function of detection consistency. For each gene list (Tier 0, Tier 0–1, Tier 2–1, Tier 3–2, Random control), the gene symbols were converted into a Boolean representation (simply connected by the operator [OR]). Each Boolean-converted list was used as a query in Pubmed and the number of hits was detected. The primary query for each gene list was modified by 4 sub-queries, from left to right: Checkered bars: [gene list] + [(disease or pathology or disorder) and (Alzheimer’s or Alzheimer or neuropathy or neurodegeneration)]; black bars: [gene list] + [cancer]; striped bars: [gene list] + [(disease or disorder or pathology)]; grey bars: [gene list] + [(disease or pathology or disorder) and stress]. The modified queries produced the numbers of hits smaller than the number of hits produced by un-delimited gene list in Boolean form. The ratios of the database responses for the modified vs. unmodified query were plotted for each group of four bars representing a consistency tier. The relative frequencies (ratios) for the queries [(disease or pathology or disorder) and stress] and [(disease or pathology or disorder) and (Alzheimer’s or Alzheimer or neuropathy or neurodegeneration)] were multiplied by 10 for convenience of representation and analysis. The Tiers 0–3 represent the lists of genes obtained as disclosed in the Methods; the Tier 1–0 is the result of subtracting the Tier 0 list from the Tier 1 list; the Tier 2–1 is the result of subtracting the Tier 1 from the Tier 2 list; the Tier 3–2 is the result of subtracting the Tier 2 from the Tier 3 list; the Random control set was obtained by randomly selecting the genes among Affymetrix and Illumina total lists and the list is not expression intensity normalized.