The MetaGraph framework is designed to work with a wide range of input data sets, indexing from a few samples up to the contents of entire archives with hundreds of thousands of records. The indexing workflow always follows the same principle, transforming single input samples into error-removed, refined sample graphs, which are then merged into a joint metagraph index. Each input sample is annotated in the joint index as a subgraph. This graph index enriched with metadata can then be used for downstream applications such as sequence search or differential assembly.

Once constructed, a MetaGraph index provides a powerful resource for data analysis. All information contained in the index can be efficiently retrieved via sequence search. The current framework offers both high throughput exact matching of k-mers, intersecting the k-mers present in the query with the k-mers in the index and returning the annotation labels of the intersection, as well as sequence-to-graph alignment, which returns all paths in the graph within a certain edit-distance and their associated annotation labels.

Differential assembly is another concept proposed in MetaGraph, which aims at finding sequences specific to certain groups of interest. For instance, these can be tissue-specific splice variants in RNA. As a special case, this can also be used for a simple large-scale de novo genome assembly.