# Duplicate long hits from PSI-BLAST

I had a protein Refseq ID and I PSI-BLASTed this sequence against Refseq database. We all know that the Refseq is a Reference sequence database and it shouldn't have redundancy. After BLASTing my sequence, at first iteration I got 1000 hits and among them there were a lot of redundant sequences!

My sequence had 241 amino acids and I found a lot of sequences with 100% identity, 100% cover and 0 E-value exactly the same as my sequence but with different IDs. All of these IDs were from RefSeq! In other iterations and after adjusting format options, I got this redundancy with other sequences from other species. My sequence is related to a chain of a multichain protein (E.coli fumarate reductase)

For example, when using NP_418578 as a query, I found WP_078165098.1, WP_064226696.1, WP_062863447.1, WP_001401474.1 and other that were identical.

I want to know what is wrong with Refseq. Is it really a Reference sequence database? If it is, where does this redundancy come from and why?

• I assume you were using the protein version of the database, right? And were all the hits from the same species? Please edit your question and show us a specific example. Show us your query sequence and a few of the redundant results. – terdon Jun 19 '17 at 17:53
• Yes, I am using the protein version of the database and my Refseq ID is: NP_418578.1 other Refseq IDs the same as this one are: WP_078165098.1, WP_064226696.1, WP_062863447.1, WP_001401474.1 and others – Sara Jun 19 '17 at 18:08
• @Sara Maybe you could post a screenshot of your results (and queries). But having a short sequence of 241aa in several sequence might indicate good conservation (maybe it is a domain?) – llrs Jun 20 '17 at 8:51
• @Sara Please edit your question to add extra information. Comments are easy to miss, hard to read and can be deleted without warning. – terdon Jun 20 '17 at 10:25

This is what is known as a feature, not a bug. Note that your identical proteins all have accessions starting with WP_. These are special "non-redundant proteins". Many sequences—particularly bacterial sequences—are identical between various different species so having a separate RefSeq entry for each of them would be inefficient. Therefore, RefSeq combines multiple such proteins into a single WP_ record. This is documented here (emphasis mine):

A new type of RefSeq protein record which represents non-redundant protein sequences was introduced in mid-2013. This record type was introduced to address a growing issue with redundancy in the Prokaryotic RefSeq protein dataset that coincided with a significant increase in bacterial genome submissions from individual isolates and closely related bacterial strains. For example, a large number of high-quality bacterial genomes may be submitted during a disease outbreak. The submitted sequences may reflect pathogen evolution during the course of the outbreak but the majority of the encoded proteins from these genomes may be identical to each other. As RefSeq includes these genomes, per community requests, this resulted in increased redundancy. By representing identical proteins using a single non-redundant protein accession number (with the prefix 'WP_'), redundancy in the database is significantly reduced.

[ . . . ]

Because a non-redundant protein sequence may be found in RefSeq genomes from multiple species, the organism information provided on the protein record reflects the lowest-common taxonomic node ranging from the genus species level to super-kingdom. A non-redundant protein record that provides organism information at the level of a genus, family, or even super-kingdom does not mean that the protein is found in all RefSeq genomes below that taxonomic classification. It only indicates that the protein is found in more than one genome of different species for which the genus, family, or super-kingdom classification is the lowest common taxonomic node.

So, your query was NP_418578.1, anaerobic fumarate reductase catalytic and NAD/flavoprotein subunit from E. coli strain K-12, substrain MG1655. The first thing to notice is how specific that is. This is the protein found from one specific substrain of one specific strain of one specific bacterial species. It is reasonable to expect that there will be identical sequences from many, many closely related species. Both from, most probably, all other strains and substrains of E. coli and from other, related bacteria.

Now, the specific sequences you mention are actually slightly different and not 100% identical. Below is a multiple alignment of NP_418578.1 and the 4 WP_ sequences you mentioned. Note that each of the 5 entries is slightly different. Each has one residue that differs from the rest. Look for the : in the identity line, there are 4 : and all others are * (I am only showing the relevant alignment blocks here and have removed those where all 4 sequences were identical):

WP_001401474.1      MQTFQADLAIVGAGGAGLRAAIAAAQANPNAKIALISKVYPMRSHTVAAEGGSAAVAQDH
*******************************************************:****

[ . . . ]

***********************************************:************

[ . . . ]

WP_001401474.1      GILMTEGCRGEGGILVNKNGYRYLQDYGMGPETPLGEPKNKYMELGPRDKVSQAFWHEWR
WP_062863447.1      GILMTEGCRGEGGILVNKNGYRYLQDYGMGPETPLGEPKNKYMELGPRDKISQAFWHEWR
WP_064226696.1      GILMTEGCRGEGGILVNKNGYRYLQDYGMGPETPLGEPKNKYMELGPRDKVSQAFWHEWR
NP_418578.1         GILMTEGCRGEGGILVNKNGYRYLQDYGMGPETPLGEPKNKYMELGPRDKVSQAFWHEWR
WP_078165098.1      GILMTEGCRGEGGILVNKNGYRYLQDYGMGPETPLGEPKNKYMELGPRDKVSQAFWHEWR
**************************************************:*********

WP_001401474.1      KGNTISTPRGDVVYLDLRHLGEKKLHERLPFICELAKAYVGIDPVKEPIPVRPTAHYTMG
WP_062863447.1      KGNTISTPRGDVVYLDLRHLGEKKLHERLPFICELAKAYVGVDPVKEPIPVRPTAHYTMG
WP_064226696.1      KGNTISTPRGDVVYLDLRHLGEKKLHERLPFICELAKAYVGVDPVKEPIPVRPTAHYTMG
NP_418578.1         KGNTISTPRGDVVYLDLRHLGEKKLHERLPFICELAKAYVGVDPVKEPIPVRPTAHYTMG
WP_078165098.1      KGNTISTPRGDVVYLDLRHLGEKKLHERLPFICELAKAYVGVDPVKEPIPVRPTAHYTMG
*****************************************:******************

[ . . . ]


Your sequence (NP_418578.1) is only identical to one WP_* multi-species sequence, WP_001192973:

WP_001192973.1      MQTFQADLAIVGAGGAGLRAAIAAAQANPNAKIALISKVYPMRSHTVAAEGGSAAVAQDH
************************************************************

WP_001192973.1      DSFEYHFHDTVAGGDWLCEQDVVDYFVHHCPTEMTQLELWGCPWSRRPDGSVNVRRFGGM
NP_418578.1         DSFEYHFHDTVAGGDWLCEQDVVDYFVHHCPTEMTQLELWGCPWSRRPDGSVNVRRFGGM
************************************************************

************************************************************

WP_001192973.1      LVQIRANAVVMATGGAGRVYRYNTNGGIVTGDGMGMALSHGVPLRDMEFVQYHPTGLPGS
NP_418578.1         LVQIRANAVVMATGGAGRVYRYNTNGGIVTGDGMGMALSHGVPLRDMEFVQYHPTGLPGS
************************************************************

WP_001192973.1      GILMTEGCRGEGGILVNKNGYRYLQDYGMGPETPLGEPKNKYMELGPRDKVSQAFWHEWR
NP_418578.1         GILMTEGCRGEGGILVNKNGYRYLQDYGMGPETPLGEPKNKYMELGPRDKVSQAFWHEWR
************************************************************

WP_001192973.1      KGNTISTPRGDVVYLDLRHLGEKKLHERLPFICELAKAYVGVDPVKEPIPVRPTAHYTMG
NP_418578.1         KGNTISTPRGDVVYLDLRHLGEKKLHERLPFICELAKAYVGVDPVKEPIPVRPTAHYTMG
************************************************************

WP_001192973.1      GIETDQNCETRIKGLFAVGECSSVGLHGANRLGSNSLAELVVFGRLAGEQATERAATAGN
NP_418578.1         GIETDQNCETRIKGLFAVGECSSVGLHGANRLGSNSLAELVVFGRLAGEQATERAATAGN
************************************************************

WP_001192973.1      GNEAAIEAQAAGVEQRLKDLVNQDGGENWAKIRDEMGLAMEEGCGIYRTPELMQKTIDKL
NP_418578.1         GNEAAIEAQAAGVEQRLKDLVNQDGGENWAKIRDEMGLAMEEGCGIYRTPELMQKTIDKL
************************************************************

WP_001192973.1      AELQERFKRVRITDTSSVFNTDLLYTIELGHGLNVAECMAHSAMARKESRGAHQRLDEGC
NP_418578.1         AELQERFKRVRITDTSSVFNTDLLYTIELGHGLNVAECMAHSAMARKESRGAHQRLDEGC
************************************************************


So, in summary, RefSeq will combine multiple identical sequences into a single WP_* multi-species accession. You should therefore expect to find one 100% identical WP_* sequence for your query and multiple, almost identical WP_* entries. And that's precisely what you see here.
• @Sara if this is still about finding homologs, then just include whatever you find. The more (relevant) sequences you include, the better your results will be. Now, if the NP and the WP sequences are 100% identical, it makes no difference at all which one of the two you use as a query since they are the same sequence. By the way, feel free to ping me (@terdon) in the Bioinformatics chat room if you need more clarifications. – terdon Jun 20 '17 at 14:33