Collections-specific Searches in Genetic Databases

These are the slides from my TDWG talk on October 5, 2017. (doi: 10.3897/tdwgproceedings.1.20320)

I did a brief overview of how specimen data is stored in GenBank records, how it can be searched, and how the NCBI LinkOut system can provide a "patch" on incorrect or incomplete specimen data.

I think introduced a tool I built called "genetic_collections", which can be found on GitHub here: https://github.com/MikeTrizna/genetic_collections.

Transcript

1. Collections-specific Searches in Genetic Databases Mike Trizna Smithsonian Institution October 5, 2017 TDWG 2017
2. Can you tell me how many GenBank records are derived from vouchers in your institution’s collection?
3. The ”official" way to find this answer • By searching on indexed records connected to NCBI’s list of institution codes
4. The ”official" answer • But… this requires specimen_voucher (or biomaterial or culture_collection) to be in properly constructed DwC triplet format
6. The "real" answer (Not indexed by GenBank, because it is separated by a space and not a colon)
7. The “real” answer
8. How to ensure these extra records show up in the proper search • Have sequence authors send NCBI updates to records
9. A "practical" solution (while hopefully working on correct solution): • Use NCBI LinkOuts to organize records that you know come from your institution
10. What is a LinkOut? LinkOut is a service that allows you to link directly from PubMed and other NCBI databases to a wide range of information and services beyond the NCBI systems.
11. LinkOut Pros • Once registered as a LinkOut provider, you can put a LinkOut on any record – regardless of ownership
12. LinkOut Pros • LinkOuts can be queried to get more accurate counts
13. LinkOut Cons They don’t exactly stand out on a GenBank record.
14. NCBI LinkOuts • Only 2 scientific collections are using LinkOuts in a significant way:
15. One more issue to bring up
16. Introducing “genetic_collections”
17. genetic_collections command line tools Written as a Python library, but installation also packages stand-alone command line tools – no Python knowledge required!
18. genetic_collections workflow Institution Search Name queries Name or code matches with counts for context Other search parameters like Taxonomy, Date Ranges, etc. List of IDs Count of Results Tabular data formats: (Excel, TSV, CSV) Database Search Database Fetch Inputs Outputs
19. Institution Searching If you’re searching by an Institution Name or Institution Code, it helps to know what names or codes are being used, and how many times they’re used. • NCBI BioCollections has 6812 institutions, – But only 2103 have at least 1 GenBank record • BOLD has 1940 unique “institution storing” values – Only 900 with 50 or more public records
20. Institution Searching
21. Database Searching
22. Database Fetching
23. Acknowledgements • Special thanks to: – Diane Pitassy, collections manager for the NMNH Fish Collection – Tom Hollowell from the NMNH Informatics Department – The SI DNA Barcoding Network and Global Genome Initiative teams