Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

From: MGcount: a total RNA-seq quantification tool to address multi-mapping and multi-overlapping alignments ambiguity in non-coding transcripts

Fig. 2

MGcount strategy. a MGcount takes a set of genomic alignments (BAM files) and a GTF RNA feature annotations file as inputs. The algorithm assigns reads hierarchically and then models multi-mapping assignments in a graph using the Rosvall’s map equation [36, 37]. As output, MGcount provides an RNA expression count matrix (where feature communities are collapsed as new defined features), a feature metadata table and the graphs. b Illustration of how the hierarchical assignation can resolve multi-overlappers: reads that map to small-RNA and long-RNA features are assigned to small-RNA in the first round; reads that map to long-RNA introns and long-RNA exons are assigned to long-RNA exons in the second round; remaining reads are assigned in the last round. c Illustration of multi-mapping small-RNA and long-RNA exon graphs generation by MGcount. Reads ri (i = 1, 11) have been hierarchically assigned to \(S_{1}, S_{2}, S_{3}, S_{4}, S_{5}\) (small-RNA biotypes, yellow), and \(G_{1}, G_{2}\) (long-RNA biotypes, blue). Each vertex in the directional multi-mapping graphs (right) corresponds to a feature and has a size proportional to the logarithm of the number of alignments. Edges connect vertices with common multi-mapping reads, with weights proportional to the number of common multi-mappers normalized by the total number of alignments of the source vertex. Hence, the weight of the edge connecting S1 with S2 becomes 3/4 (reads mapping both S1 and S2 divided by reads aligned to S1). (CB: Cell Barcode, UMI: Unique Molecular Identifier)

Back to article page