ncbi’s bioinformatics resources michele r. tennant, ph.d., m.l.i.s. health science center...
TRANSCRIPT
NCBI’s Bioinformatics NCBI’s Bioinformatics ResourcesResources
Michele R. Tennant, Ph.D., M.L.I.S.
Health Science Center Libraries
U.F. Genetics Institute
January 2015
Entrez Nucleotides
Entrez Nucleotides (GenBank)
• Database of nucleotide sequences (ATGC)
• Actually contains data from several databases - GenBank, EMBL, DDBJ, RefSeq
• Hard to search because many submitting scientists send in redundant information and poorly annotated information
Nucleotide Data DomainNucleotide Data Domain
• As of December 15, 2014
• Over 184,938,063,614 bases
• Over 179,295,769 sequence records
• Some complete genomes and chromosomes
So Why So Hard to Search?So Why So Hard to Search?
• No controlled vocabulary - lose power of MeSH - must OR synonyms. Often miss the records you want.
• Archival - quality of annotations depends on the submitter (especially features field); little to no quality control; spelling errors! Often miss the records you want.
• Redundant - lots of records for the same gene; partial records, etc. Often pull up records you don’t want.
GenBank Sample Record
• Before searching, we will look at a GenBank sample record
• Note that the “Features” field provides useful biological information, and may be searched
Click any link in sample record
to access definition of
field and search tips
“Definition” field acts as record title – search
[titl]
Unique identifier; assigned by NCBI;
required by journals/grants
Link to PubMed citation/abstract
The “Features” field provides the most
biological information; search as [fkey]
Numbers indicate
location on the nucleotide sequence
…3158
GenBank IdentifiersGenBank Identifiers
• Accession Number - U49845 [accn]• Unique identifier; does not change• Letter prefix no longer has significance
• Version - U49845.1 • If any change to sequence, version
U49845.2 created• GenInfo Identifier (GI number) [uid]
• Run parallel to accession.version system; change in sequence changes number
Searching “Nucleotides”• Database is difficult to search:
• Redundant records• Archival - poor or missing annotation
• Best searches are done using commands; need a class to learn all
• Practice search – search for sequences for human presenilin 1• Is there anything odd about the some of the
retrieved results?
Search for HUMAN presenilin 1
But end up with rat, mouse, etc.
Choose “nucleotide” from dropdown, then
click “search”
Searching “Nucleotides”• We retrieved the non-human and PSEN2
(rather than PSEN1) records because the computer looked for the terms “human” and “presenilin 1” ANYWHERE in the record (click on details tab to see how the computer parsed your search)
• Use complex boolean searching to clean this up: term [field] AND term [field]
Searching “Nucleotides”• How to get rid of non-human sequences?
• Search human [orgn] (this works for any taxon)
• How to get rid of non-presenilin 1 sequences?• Another trick – search PSEN1 [gene]• Note – you may miss relevant sequences, but should not pick
up irrelevant sequences• The sequences that you miss are the ones that have not been
annotated with the current official gene symbol in the “gene” field
• DO NOT use this method if you need to find every sequence for a particular gene
• Human [orgn] AND PSEN1 [gene]
Use these filters to choose molecule type, confine to
RefSeq records
This is the search that was completed using fields (orgn, gene) and filters
How Can I Find “Best” Sequences
• Non-redundant, curated subset of the sequence data domains
• Contains one record for each gene or splice variant from each organism represented
• Records can be thought of as “review articles” for sequences
• “Best” (usually longest) sequence used as seed• Value-added annotations provided by experts• Easy – a tab now exists to limit retrieval to just RefSeq
Click on the RefSeq link to retrieve only the “best” sequences (highly
annotated, complete, nonredundant)
The typical RefSeq accession number
format: 2 letters, an underscore, and
then numbers
Viewing Formats
• The “Default” view is the standard GenBank record
• Researchers often use the “FASTA” format for analysis
• Change the record format at the “Display” pull-down menu
Entrez Proteins
Entrez ProteinsEntrez Proteins
• Contains data from several databases:
• SwissProt, PIR, PRF, PDB
• Translations from annotated coding regions in GenBank and RefSeq
• Redundant archival data domain of publicly available protein sequences
Searching Entrez ProteinsSearching Entrez Proteins
• Searched like Entrez Nucleotides• “Filters” choices differ; includes
molecular weight and sequence length filters
Entrez Gene
Entrez Gene
• Pulls together information (sequences, structures, literature, gene models, pathways, etc.) for genes
• Best place to start for “gene-centered” info• One record per gene per organism• Search by names, symbols, accessions,
publications, GO terms, chromosome numbers, E.C. numbers, etc.
Search using gene symbol
Could have searched under any of these
aliases (unlike GenBank where you
would have to try them all)
Official gene symbol as
determined by the Human
Genome Nomenclature Commission
Summary of protein, function
and disease-causing mutations; from RefSeq record
Links to PubMed records
that provide evidence of
function – any researcher can
add these
Links to OMIM records of phenotype/
disease
Gene Ontology terms form a controlled vocabulary with three components – biological process, molecular function, and cellular
component
Links to homology
maps
Links to protein interactions
Pathway info may be available from
the Kyoto Encyclopedia of
Genes and Genomes
Sequence and domain links
Links to GeneReviews – clinical resource
Taxonomy Browser
Search Taxonomy BrowserSearch Taxonomy Browser
• How many genera from the family Iguanidae are represented by sequence data?
• How many nucleotide and protein sequences are available for the family?
Entrez Searching Summary
To Find Everything(?) To Find Everything(?) Broaden SearchBroaden Search
• OR together synonyms • OR together related terms (gene name, gene symbol,
protein name, alternate spellings, disorder)• Don’t specify a field- search entire record• Truncation - use * at end of word root• Click “Related Records”• Try using Taxonomy Browser to pick up all taxa in a
particular group
Fewer/Best Records Fewer/Best Records Narrow SearchNarrow Search
• Search particular fields:
• PubMed - MeSH Browser, subheadings, major MeSH
• Nucleotide - features, title, gene, properties, organism
• Use “Filters”• Search only the RefSeq database
Will Entrez Find Will Entrez Find Every Sequence Record?Every Sequence Record?
• No!!! • Entrez relies on annotation of records, so you are
searching solely on “terminology”• Some records are not annotated, some records are
poorly or incorrectly annotated
• To find all useful sequences – need to search on sequence itself • Related sequence link• BLAST
Entrez “Related Records”Entrez “Related Records”
• Will vary depending on data domain• PubMed related articles
• Based on a “word weight” algorithm – MeSH, title, abstract words
• In order by weight (highest weight first)
• Nucleotide and protein related sequences• Based on basic BLAST search• In order by best BLAST score