Part I: What is MiGA?
MiGA organizes your data in a consistent, well-organized fashion independent of centralized databases. This makes MiGA projects the ideal system to store data even if you don't use MiGA for anything else. MiGA is completely based on filesystem structures, so it can easily be transferred, backed-up, and stored long-term. Moreover, MiGA projects can be easily browsed, with descriptive folder names and a simple structure that is easy to understand.
MiGA is not designed to support versioning or database storage, other than individual file-based databases, in order to keep the overhead on any of the tasks above (and the system requirements) at a minimum.
MiGA performs general-purpose analyses to pre-process genomic and metagenomic data. The main purpose of MiGA is genome-based taxonomy, but some pre-processing steps are necessary regardless, so they can be used for many other purposes. For example, the initial data in most genomic and metagenomic projects is sequencing data. For almost any project, this means that trimming, clipping, and read-quality assessment are necessary steps for any downstream analyses. In most cases, assembly and gene prediction are also necessary, and other analyses like rRNA and essential genes detection is very useful. All of this is automatically done by MiGA!
MiGA is not a workflow manager system. MiGA only supports short-read data (and it's optimized for Illumina data) or already assembled sequences. MiGA's goal is to keep analyses as simple and standardized as possible, so only critical customization is supported.
MiGA is designed to process genomes and can handle metagenomes (with some restrictions). MiGA is optimized for short-read datasets or assembled datasets. MiGA is optimized to process prokaryotic data (Archaeal and Bacterial), but it has some readily available customizations for viral metagenomes (or viromes). For more details, see the types of datasets and projects and the input data supported.
MiGA does not have custom settings for eukaryotic or viral genomes, nor for transcriptomic data. The data management design (and perhaps some of the processing steps) can be used for these and other purposes, but thread carefully.
MiGA has a general-purpose design with some presets designed for the different data types supported. All internal configuration and metadata are stored as individual JSON files. Sequence quality is stored as FastQ, and sequences are stored as FastA; these two cover most of the data in the system. There are also some graphic reports in PDF and HTML, raw-text reports and logs, and general statistics in JSON. Finally, all of the pair-wise comparisons are stored in SQLite3 files described here.
- daemon/: Daemons lair.
- daemon.json: Daemon settings.
- ...: Several daemon log files.
- data/: All the data is stored here.
- 07.annotation/: Data annotations.
- 08.mapping/: (Currently not in use).
- 09.distances/: Pair-wise comparisons(distances).
- 04.ssu/: (Currently not in use).
- 10.clades/: Dataset clustering at various resolution levels.
- 04.phylogeny/: (Currently not in use).
- 05.metadata/: (Currently not in use).
- metadata/: Collection of JSON files with datasets metadata.
- miga.project.json: JSON file with project metadata.
MiGA's ultimate goal is to provide a standardized set of tools for consistent genome-wide taxonomic analyses. For this reason, MiGA does not provide nor favor any one taxonomic database. This authority-agnostic approach allows us to focus on the underlying analyses, supporting as many schemas as possible. With that being said, MiGA does support automated taxonomy annotation for some databases in EBI and NCBI linked to NCBI Taxonomy, and it does support some automated adjustments for the JGI schema (in particular for metagenomes). Instead of forcing groups by external taxonomies that may have varying degrees of accuracy and completeness, MiGA follows a data-driven clustering based on naturally-forming groups based on AAI and ANI analyses. Hence, MiGA projects can be used to classify novel genomes using any reference taxonomy (or none!).
MiGA does not provide or endorse any particular taxonomic authority.
MiGA can catalogue datasets, even in the absence of a reference taxonomy. This allows many advanced analyses, including (but not restricted to):