I have more than 100 sequences which I want to BLAST against UniProt database. However, the web version of the BLAST (https://www.uniprot.org/blast) limits the number of sequences to only 5. In such a case, is there a programmatic way to BLAST the UniProt database or other approach to BLAST many sequences at once?
2 answers
For any number of query sequences, if you have room in your hardrive and you are willing to go command-line, you can download UniRef FASTA files from https://ftp.uniprot.org/pub/databases/uniprot/current_release/uniref, format them with makeblastdb and run BLASTP/BLASTX against them with something like:
makeblastdb -in uniref50.fasta -dbtype prot
blastp -query input.fasta -db /path/to/uniref50.fasta
Note that UniRef sets contain clusters of UniProt and Uniparc sequences, read more at https://www.uniprot.org/help/uniref. They are available with 100%, 90% or 50% redundancy cutoffs, and the corresponding compressed FASTA files take 99GB, 43GB and 12GB respectively.
Sequences within a UniRef50 cluster share at least 50% sequence identity with each other. They group together a broader range of protein sequences, including more distant homologs. UniRef100 clusters contain sequences with a sequence identity of 100%. These clusters are more focused and tend to represent closely related sequences. Hence, the ouput will change depending on which cluster you are using.
Will there be a difference in final output (E-score, Percent Identity, Query cover) if UniRef50 is used instead of UniRef100
Yes. Since the database content is going to be different as shown by the size differences noted above. Blast depends on the database contents to generate statistics and alignments.
@b.contreras.moreira Uncompressed UniRef100 FASTA file size is 943 GB, which is way beyond my local computer storage and uncompressed UniRef50 FASTA file is 24.4 GB. I realized that my sequences are all supposed to be Human proteins and therefore I downloaded the human proteome (Swiss-Prot + TrEMBL) from UniProt and ran BLAST locally, which is more efficient way.
If you cannot use local BLAST as suggested above (which may be an excellent alternative, be it with UniProtKB, UniRef or just UniProtKB/Swiss-Prot), UniProt recommends programmatic access as described in this EBI help page: https://ebi-biows.gitdocs.ebi.ac.uk/documentation/webservices/
Go to the "Sequence similarity search" section and select NCBI BLAST+.
Log in to answer this question.
-
Optimum setting for local blastp for ~10K sequences
written by sodiumnitrate •
-
What does Uncharacterized Protein Uniprot designation mean?
written by emalekos •
-
Blastn for gRNA characterization in ANopheles
written by Sbrillo •
-
Blastp against CAZy database
written by huiyus97 •
-
The best way to count a number of insertion sequences in a set of bacterial genomes
written by little_more •
-
Annotate coding sequences
written by karthic
-
Are COG IDs consistent across different versions of the eggNOG database?
written by jjoh090 •
-
Dealing with UniProt humsavar database reference ambiguity
written by Eli Korvigo
-
Downloading Human And Other Completely Sequenced Proteomes To Search Homologs
written by Pappu
-
How To Find Sequences For Protein Names (A Challenge)
written by Simon Cockell
If you are willing to forgo
tremblpart then you could useswissprotat NCBI protein web blast.Thanks! This is a fast way to check if manually annotated entries are enough as database.