Deep learning is an increasingly used tool in biology and biophysics. From predicting protein structure, to microscopy image segmentation, to recognizing motifs in the genome, deep learning is becoming a widespread, effective method for tackling many different problems in biology. One such problem is DNA regulatory sequence design - such as designing promoter regions - for controlling gene expression. Recently, generative models based on deep learning have been able to design novel regulatory sequences for E. coli (and separately for S. cerevisiae) leading to gene expression levels exceeding native genes. In order to do so, generative adversarial networks are used for generating novel sequences, while convolutional neural networks are used for predicting gene expression levels. The gut microbe B. theta has been difficult to engineer. Because B. theta is prominent in the Western diet, it is becoming a candidate chassis organism for potentially engineering biosensors and therapeutics. Given B. theta's importance, I discuss preliminary results from using a deep learning model to predict the gene expression of given regulatory sequences for the gut microbe B. theta.