In order to train the deep learning model, we need to prepare the supervised training data first. in the causalcall(doi: 10.3389/fgene.2019.01332),to effectively train the network, sequencing data are labeled in the workflow of basecalling, re-squiggle, and segmentation. First, a basecaller is used to identify the base sequence of the sequencing read. Then, the re-squiggle process is performed, involving two steps. In the first step, a base sequence is mapped to the reference genome to obtain the correct sequence. In the second step, the correct sequence is mapped back to the current measurements. The above workflow is nice and worked very well. But if we have no the first basecaller, in the very beginning, how the raw current signal map to base without a basecaller?
