Identifying Colorectal Tumor For Single Cell RNA Sequence Using Rectified Linear Unit With Stochastic Gradient Descent

Abstract

To detect cell clusters in whole human colorectal tumor cells, mechanisms based on single-cell RNA sequencing have been reported. To address such issues, a deep learning technique for single-cell analysis was recently developed, with promising results. Single- cell RNA sequence data contains detailed information about the transcriptome profile. Through gene expression, this information is presented in the form of patterns. The neural network is shown here to help understand these data representatives. Then we use the differential expression to identify distinct cell types and gene markers. Finally, the proposed identifying colorectal tumor for single-cell RNA sequencing that Rectified Linear Unit (ReLu) with Stochastic Gradient Descent(SGD) is compared to other recently developed models such as ReLu with Adam, ReLu with Limited-BFGS-B, TanH with Adam, TanH with SGD, TanH with Limited-BFGS-B, Sigmoid with Adam, Sigmoid with SGD, Sigmoid with Limited-BFGS-B, Linear with Adam, Linear with Limited-BFGS-B and Linear with SGD. At the moment, the results from identifying colorectal tumor for single-cell RNA sequencing that ReLu with SGD performs better than other recently developed models.