Flowchart of TransMeta. (A) Building individual splicing graphs for each sample in parallel. The nodes (exons) are named with lowercase letters; the values on each edge are the weights of the edge. (B) Merging individual splicing graphs into the vector-weighted splicing graph (VWSG). The nodes and edges in the VWSG are all the nodes and edges appearing in all the samples. Each edge in the VWSG is weighted by a vector, with the kth component of the vector being the corresponding weight of the edge in the splicing graph of sample k. (C) Using the cosine similarity-based combing strategy and the label-setting technique to generate the initial meta-assembly. (D) Extracting assembly for each sample from the VWSG. (E) Producing the mature meta-assembly.
