Author/Authors :
r Nosirov, Bakhtiyo Systems Immunology Lab - Osaka University, Suita , Billaud, Joël Systems Immunology Lab - Osaka University, Suita , Vandenbon, Alexis Systems Immunology Lab - Osaka University, Suita , Diez, Diego Systems Immunology Lab - Osaka University, Suita , Wijaya, Edward Systems Immunology Lab - Osaka University, Suita , Ishii, Ken J Systems Immunology Lab - Osaka University, Suita , Teraguchi, Shunsuke Systems Immunology Lab - Osaka University, Suita , Standley, Daron M Systems Immunology Lab - Osaka University, Suita
Abstract :
Purpose: Evidence suggests that circulating serum microRNAs (miRNAs) might preferentially
target immune-related mRNAs. If this were the case, we hypothesized that immune-related
mRNAs would have more predicted serum miRNA binding sites than other mRNAs and,
reciprocally, that serum miRNAs would have more immune-related mRNA targets than non-
serum miRNAs.
Materials and methods: We developed a consensus target predictor using the random for-
est framework and calculated the number of predicted miRNA–mRNA interactions in various
subsets of miRNAs (serum, non-serum) and mRNAs (immune related, nonimmune related).
Results: Immune-related mRNAs were predicted to be targeted by serum miRNA more than
other mRNAs. Moreover, serum miRNAs were predicted to target many more immune-related
mRNA targets than non-serum miRNAs; however, these two biases in immune-related mRNAs
and serum miRNAs appear to be completely independent.
Conclusion: Immune-related mRNAs have more miRNA binding sites in general, not just for
serum miRNAs; likewise, serum miRNAs target many more mRNAs than non-serum miRNAs
overall, regardless of whether they are immune related or not. Nevertheless, these two independent
phenomena result in a significantly larger number of predicted serum miRNA–immune mRNA
interactions than would be expected by chance.
Keywords :
biomarker , posttranscriptional regulation , random forest , target prediction