State-of-the-art structural variant calling: What went conceptually wrong and how to fix it?
Schmidt, M. R.; Kutzner, A.
Show abstract
Structural variant (SV) calling belongs to the standard tools of modern bioinformatics for identifying and describing alterations in genomes. Initially, this work presents several complex genomic rearrangements that reveal conceptual ambiguities inherent to the SV representations of state-of-the-art SV callers. We contextualize these ambiguities theoretically as well as practically and propose a graph-based approach for resolving them. Our graph model unifies both genomic strands by using the concept of skew-symmetry; it supports graph genomes in general and pan genomes in specific. Instances of our model are inferred directly from seeds instead of the commonly used alignments that conflict with various types of SV as reported here. For yeast genomes, we practically compute adjacency matrices of our graph model and demonstrate that they provide highly accurate descriptions of one genome in terms of another. An open-source prototype implementation of our approach is available under the MIT license at https://github.com/ITBE-Lab/MA.
Matching journals
The top 4 journals account for 50% of the predicted probability mass.
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.