BioAnnotate

Annotation-layer for the INSDC BioSample database

Motivation

The BioSample database contains descriptive meta-data for all biological samples housed by the "International Nucleotide Sequence Database Collaboration", the world's central repository for biological sequence data.

Due to the diversity of available samples and their descriptions, the meta-data is not standardized. Each record is stored as an XML file, containing its own set of tags and values.

BioAnnotate provides an annotation layer for BioSample to aggregate similar tags and standardize the value formats.

Goal

There are 42,125 unique 'tags' across >10.7 million BioSample XML files. We will annotate these tags into 4 categories, to allow for data-aggregation and ultimately a "clean" database.

geo: Geographic names and spatial coordinates
date: Sample collection date and/or release date
organism: Host or Pathogen species
ecosystem: Environmental origin description or body-site

Contributing

BioAnnotation Spreadsheet

We will work on a collaborative Annotation Spreadsheet which contains every unique BioSample tag.

Sign-up on the Lockout sheet to annotate a 'chunk' of 2,500 rows in the biosample_tags sheet for a particular class of data (see below).

The default for all tags is set to F for "FALSE". If a biosample_tag describes a field which is pertinent to your data-class, change this value to T for "TRUE".

If you are unsure of how to classify a particular biosample_tag, set the value to ? and/or ask in the chat.

Example Workflow

Kat would like to annotate Chunk C for geo data.

She reviews the geo data class description below to understand the inclusion and exclusion criteria for this data-class.
She enters her name on the Lockout sheet to indicate she has begun to work on this chunk.
On the biosample_tags sheet Chunk corresponds to Rows 5001 - 7500 and the geo_name and geo_coord columns.
After turning on some good jams, she annotates these rows.
Upon completing her annotation, she updates Lockout to indicate this chunk is complete and she can begin working on another Chunk.

Data Classes

geo data

Inclusion: Tags which can provide any location data. Imagine the keywords you could type into Google Maps. e.g. geo_location, country, national_park, sequencing_institute, lake_name, longitude, lat_long, geo_coordinates...

Exclusion: Tags which describe a generic environment, not geographically specific. e.g. snow_depth, nitrogen_content_soil, lake_type...

geo_name : Set to T if tag likely contains words describing geo-data.
geo_coord : Set to T if tag likely contains numbers describing geo-data, mainly longitude / latitude / altitude.

date data

Inclusion: Tags which would contain a date. e.g. collection_date, sample_date, sequencing_date, release_date...

Exclusion: Tags which contain time-course data, such as the timeline of an experiment. e.g. week_of_growth, hours...

collection_date : Set to T if tag specifically describes the time at which a sample was collected from nature.
other_date : Set to T if tag contains a date.

organism data

Inclusion: Tags which can provide taxonomic information regarding the organism which had been sampled. e.g. species, genus, scientific_order, taxonomy_string

Exclusion: Tags which describe a generic component of an organism. e.g. leaf_type, fur_colour, paw_length...

host_species : Default choice to set to T for this class
virus_species : Set to T if tag specifically indicates a viral organism classification.

ecosystem data

Inclusion: Tags which can provide an environmental or organism-tissue description of the samples origin. e.g. water_depth, wastewater_site, soil_moisture, brain_region, tumour_diameter, organ_site...

ecosystem : Set to T if tag describe the samples environment.
bodysite : Set to T if tag describes an organism's site.

Building Knowledge

An example use-case for this data showing the spatial-distribution of several million DNA or RNA sequencing datasets in the 'Sequence Read Archive'. Geographic data was extracted from BioSample, but >50% of the data is missing due to inconsitent naming, we're going to fix that!

Contributors

@ababaian @adrianbele @cbenon @linzzasaurus @mamurak @rgodinezp @schen1 @shiwanibiradar

ababaian / BioAnnotate