Analysis of transcription control signals using artificial neural networks

The role of the upstream region in controlling the transcription efficiency of a gene is well established. However, the question of predicting the extent of gene expressed given the upstream region has so far remained unresolved. Using an artificial neural network (ANN) to capture the internal representation associated with the transcription control signal, the present work predicts the rate of mRNA synthesis based on the pattern contained in the upstream region. Further, the model has been used to predict the transcription efficiency for all possible single base mutations associated with the beta-globin promoter. The simulation results reveal that apart from the experimental observation that alpha-79G-A and -78G-A mutation increases the efficiency of transcription, mutation in these regions by C or T also causes an increase in transcription. Furthermore the simulation results verify that mutations in the conserved region, in general, decrease the transcriptional efficiency. However, the results also show that certain sequence elements, when mutated, either cause a marginal increase in the level of transcription or have no effect on transcription levels. The simulation results can be used as a guide in designing mutation experiments since an a priori estimate of the possible outcome of a mutation can be obtained.