Tuberculosis (TB) continues to be depressingly persistent as a global pandemic, killing more than any other infectious agent (1), and the mechanisms underlying pulmonary pathology remain elusive (2). Diverse clinical phenomena and experimental observations recently led us to hypothesize that an infectioninitiated autoimmune phenomenon is part of the pathological process (3), and others have since noted this association (4, 5). To investigate this hypothesis, we performed an analysis of differentially expressed genes (DEGs) in patients with TB and patients with autoimmune or infectious diseases, using within-study control groups. We obtained gene expression datasets from the Gene Expression Omnibus (accession numbers GSE19444 and GSE22098, Illumina platform) to reanalyze a major published study (6). All members of the TB cohort had culture-confirmed pulmonary disease. The autoimmune diseases studied were adult and pediatric systemic lupus erythematosus, and adult Still’s disease. The infectious diseases were culture-confirmed staphylococcal or streptococcal bacteremia. All samples from patients with TB were taken before treatment was initiated. No information was available to the authors regarding the treatment status of the autoimmune cases, which could potentially introduce confounding factors. To control for variability between different microarray experiments, we first defined all DEGs in each condition relative to the healthy control subjects for each dataset. DEGs were identified using the limma package in R on quantile normalized raw expression data obtained from the Gene Expression Omnibus datasets using a moderated t test statistical filter (Benjamini-Hochberg [false discovery rate (FDR)]-adjusted P, 0.05) to define differential expression. Our analysis identified 2,468 DEGs in TB, 8,134 DEGs in infection, and 11,348 DEGs in autoimmune disorders relative to controls, resulting in disease-specific gene signatures (Figure 1). A total of 1,481 DEGs were common to all three conditions, thus representing a generic inflammatory response, and comprised 60.0% of the TB signature. Only 96 DEGs were shared exclusively between TB and infection (3.8% of the TB signature), whereas TB and autoimmune disease had 810 DEGs in common (32.8% of the TB signature). Notably, once the genes shared by infectious and autoimmune diseases were accounted for, only 81 (3.3% of total DEGs) were exclusive to TB. We further refined our disease DEG lists by quantitative filtering (fold change. 2 or, 22), which reduced the number of genes identified, and the observed disease-overlap relationships were broadly maintained (Figure 1, values in parentheses). By this more stringent analysis, 48.0% of the total TB signature was common to the autoimmune signature, whereas only 3.9% genes were common to the infectious signature. All of the 35 common genes differentially expressed in both TB and the autoimmune conditions were found to be modulated in the same direction (ie, up-or down-regulated) between TB and autoimmunity. Next, we studied the pathways that were communal between TB and autoimmune diseases. We used ToppFun functional enrichment in the ToppGene suite to determine gene ontology for disease associations for the TB–autoimmune overlap genes. Using the FDR-corrected P value–filtered common TB-autoimmune DEG list, we annotated significant (FDR P value, 0.05) associations with a number of autoimmune diseases, such as systemic lupus erythematosus, Behçet’s disease, and rheumatoid arthritis. After an additional filter for fold change (fold change. 2 or, 22) was applied, the 35 common TB …