Transcriptomic Signatures of Substance Use Disorders

We considered all the SNPs located in each DA and 5-HT gene to infer whether the genetically-predicted expression of each DA and 5-HT gene correlates with the GWAS data of the 11 phenotypes of this study (Table 1). These analyses were carried out on MetaXcan [S-PrediXcan (56–58 (link)) and S-MultiXcan (59 (link))] using the summary statistics of each disorder or trait. Prediction elastic-net models were downloaded from PredictDB,⁸ which were constructed considering SNPs located within 1 Mb upstream of the transcription start site and 1 Mb downstream of the transcription end site of each gene and were trained with RNA-Seq data of 13 GTEx (release V8) brain regions: amygdala, anterior cingulate cortex BA24, caudate, cerebellar hemisphere, cerebellum, cortex, frontal cortex BA9, hippocampus, hypothalamus, nucleus accumbens, putamen, spinal cord and substancia nigra. S-PrediXcan was used to analyse the genetically determined expression of genes in each of the 13 brain tissues described above for each of the 11 phenotypes previously selected (Table 1): alcohol dependence (45 (link)), cannabis use disorder (46 (link)), cocaine dependence (47 (link)), opioids use disorder (48 (link)), substance use disorders (49 (link)), antisocial behavior (50 (link)), disruptive behavior disorders comorbid with ADHD (51 (link)), anxiety, irritability, neuroticism, and risk taking behavior. Then, the information across tissues was combined for each phenotype using a multivariate regression with S-MultiXcan, and a multiple-testing FDR correction (5% FDR) was applied for each phenotype considering all the computed genes tested in the analyses.