The Hyphal Tip

Report concludes that a fungal genome database is of “the highest priority”.

This is the title as listed in PubMed for this article from Future Medicine about the AAM report on charting future needs and avenues of research on the fungal kingdom.

The need for a comprehensive database for information about fungi, starting at least with systematic collections of genomic and transcript data, is highlighted as a major need.  Really and sort of new database effort should strive to be more comprehensive and include genetic and population data (alleles, strains) and information like protein-protein, protein-nucleic acid interactions (as Pedro mentioned). But on top of that it, it needs to be comparative so that information from systems that serve as great models can be transferred to other fungal systems that are being studied for their role as pathogens or interacting in the environmental.

Affordable next-gen sequencing will allow us to obtain genome and transcript sequence for basically all species or strains of interest.  Researchers with no bioinformatics support in their lab will likely be able to outsource this to a company or campus core facility.  But how can they easily map in the collective information about genes, proteins, and pathways onto this new data?  And have it be a dynamic system that can update as new information is published and curated in other systems.

I think this has to be the future beyond setting up a SGD, CGD, etc for every system.  The individual databases are useful for a large enough community where there are curators (and funding), but we will have to move to a more modular system in the future (aspects of which are in GMOD) that can have both an individual focus on a specific species/clade and a more comprehensive view of the that is comparable across the kingdom.  There are 100+ fungal genomes, but the community size for some of them are in the dozens of labs or less. How can they take advantage of the new resources without an existing infrastructure of curators?  Their systems serve an important need in a research aim, but how can discoveries there make its way back into the datastream of othe systems?

I see it as there are several ways one would interact with a system that provided single-genome tools as well as a framework for comparative information.  At a gene level, one might be looking for all information about a specific gene, based on sequence similarity searches, or starting with a cloned gene in one species. Something akin to Phylofacts or precomputed Orthogroups for defining a Gene but with more linking information about function by linking in information from all sources.  So a comparative resource, but also tapping into curated andliterature mined data.

At a genome level, one might want to do whole genome comparisons of gene content from evolutionarily defined families genes (gene family size change) or at a functional level.  To start out with, each gene/protein would already need a systematic functional mapping.  This could be as simple as running InterProScan on every protein, expanded to find Orthogroups (or OrthoMCL orthologs) and transfer function from model systems, and finally even more advanced, do further classified better with tools like SIFTER.

Interlinked with these orthologous and paralogous gene sets would be anchors for analyses of chromosomal synteny and even comparative assembly including tools like Mercator.  Certainly things like all of this exist but making it more pluggable for different sets of species would be an important additional component.

At a utility level, the gene annotation and functional mapping of all this information should be possible. I would imagine a researcher could upload the sequence assembly they received from the core facility and the system can generate multiple gene predictions, annotate the genes, and link these genes within the known orthogroups of the system (preserving their privacy for these genes if desired).  Presumably this sort of thing would be easier as a standalone in-house for the researcher, but web services could also be the place for this.

For fungal-sized genomes this amount of data is not too extereme.  Things like Genome Browser, BLAST, etc should all be rolled out of the box based on the basic builds.

On the DIY and community annotation front, there would also need to be a layer of community derived annotation that could be layered on all these systems.  I would imagine this both to be for gene structure annotation (genome annotation) and functional annotation (protein X does Y based on experiment Z, here is the journal reference).  I think aspects of this would be visible, auditable (tracked), but maybe not blessed as official until a curator could oversee these inputs. In my mind, whether or not this is in a Wiki per se or just new system that allows community input is less important to me than having it be a) structured (not a bunch of free text) b) tracked and versionable c) easy for researchers to input so that the knowledge is captured, even if it has to be reorganized later on.

Seems like a lot of work to be done, but really many of these things already exist through what  the GMOD project has built.  Many loose ends and software that doesn’t fully meet up to these needs, but I think the important concept is these are all general solutions that will be of benefit to most communities, not just the fungal ones.  One lingering question I always have when approaching genomic datas

that will be dynamic, what if any of this makes its way into GenBank?  How is this sort of thing banked so that it can be captured, and does the improved functional or gene structure annotation ever make its way into the repository databases to correct and improve what has already been submitted there?