Micro-RNA Profile Detection via Factor Graphs

Recent studies have shown that micro-RNAs (miRNAs) play a key role in inter-cell communication in humans. More importantly, irregular patterns over specific miRNAs have been linked to certain types of cancer and cardiac diseases. In this paper, we introduce a general framework to sense environmental miRNAs and detect certain irregular patterns. We use a sensor cell (i.e., biosensor) array comprising of various genes whose expression can be suppressed through miRNAs of interest. Interference and noise are major issues in miRNA sensing via such a cell array. In particular, every miRNA may have a footprint on multiple biosensors and each biosensor in the array may be affected by multiple miRNAs. We present a probabilistic model capturing this phenomenon and solve the detection problem via a factor graph. Since, the exact values of the input miRNAs are not needed, fewer observation are required to achieve the same level of pattern-detection accuracy relative to directly measuring the concentration. Finally, we use Belief Propagation, a message-passing algorithm, to infer the presence of irregular patterns. Our model-based data suggests significant improvement in performance.