SEA-PHAGES | RefSeq and INSDC name disagreements in NCBI Blast for Functonal Assignment

Link to this post \| posted 17 Feb, 2023 19:50
kmaclea	I start out this post with the caveat that I may be doing something really dumb here, so I apologize if this is a known issue and I have just somehow missed it. Here's the situation: We're annotating MulchSalad (F) and had just started to teach function calling. We picked gene 1 to demonstrate with (big mistake! ). Basically here's the issue: If you BLAST on Phagesdb this gene hits to terminase small subunit. This is shown in the Pham view for other annotated genomes. If we BLAST (blastp) on NCBI with default settings, we see "minor tail protein" for the annotation: https://capture.dropbox.com/ZrKNG9p6b1vT2pDa I was confused by this and dug deeper. Apparently, this is because RefSeq is defaulting for matches. INSDC regular GenBank entries are still there with the names given by SEA-PHAGES, but they are hidden in the hits by default. You can directly compare this if you click around: https://capture.dropbox.com/u8NpcBuUFOAcg2UT If you look at the RefSeq entry it gives "minor tail protein" and if you look at GenBank it says "terminase small subunit" I am not sure if this is a common problem, but we definitely found it here. I am not sure there is really a question here so much as an observation. If I do have a question it's likely unanswerable–why did RefSeq call this MTP? HHPred agrees this is a terminase large subunit or terminase. Is this a common issue? This is the first time I'd ever seen anything like this. If it's isolated, I can deal with it ad hoc… if it's common maybe I need to plan a workaround. Thanks, all! Kyle – Kyle MacLea Associate Professor, University of New Hampshire at Manchester kyle.maclea@unh.edu +1 603-641-4129

Link to this post | posted 17 Feb, 2023 19:50

I start out this post with the caveat that I may be doing something really dumb here, so I apologize if this is a known issue and I have just somehow missed it.

Here's the situation: We're annotating MulchSalad (F) and had just started to teach function calling.

We picked gene 1 to demonstrate with (big mistake! smile

).

Basically here's the issue: If you BLAST on Phagesdb this gene hits to terminase small subunit. This is shown in the Pham view for other annotated genomes.

If we BLAST (blastp) on NCBI with default settings, we see "minor tail protein" for the annotation:
https://capture.dropbox.com/ZrKNG9p6b1vT2pDa

I was confused by this and dug deeper. Apparently, this is because RefSeq is defaulting for matches. INSDC regular GenBank entries are still there with the names given by SEA-PHAGES, but they are hidden in the hits by default. You can directly compare this if you click around:
https://capture.dropbox.com/u8NpcBuUFOAcg2UT

If you look at the RefSeq entry it gives "minor tail protein" and if you look at GenBank it says "terminase small subunit"

I am not sure if this is a common problem, but we definitely found it here. I am not sure there is really a question here so much as an observation. If I *do* have a question it's likely unanswerable–why did RefSeq call this MTP? HHPred agrees this is a terminase large subunit or terminase.

Is this a common issue? This is the first time I'd ever seen anything like this. If it's isolated, I can deal with it ad hoc… if it's common maybe I need to plan a workaround.

Thanks, all!
Kyle

–
Kyle MacLea
Associate Professor, University of New Hampshire at Manchester
kyle.maclea@unh.edu +1 603-641-4129

Link to this post \| posted 17 Feb, 2023 20:24
debbie	Hi Kyle, Well, this is quite messed up, isn't it. I will investigate further. in the meantime, I would like to provide what I think of BLASTp functions calls at NCBI, i don't value them very much - not without supporting evidence. So if you continue to investigate, there is no supporting data for a minor tail protein except NCBI said so. there is no way that a functional call can be made from the blast data that HHPred data sources does not support. (HHPred does a Psi blast, so it is finding more distant relationships to a protein than a single blast could.) Looks like this is a terminase, small subunit to me. debbie

Link to this post | posted 17 Feb, 2023 20:24

debbie

Hi Kyle,
Well, this is quite messed up, isn't it.
I will investigate further.
in the meantime, I would like to provide what I think of BLASTp functions calls at NCBI, i don't value them very much - not without supporting evidence. So if you continue to investigate, there is no supporting data for a minor tail protein except NCBI said so. there is no way that a functional call can be made from the blast data that HHPred data sources does not support. (HHPred does a Psi blast, so it is finding more distant relationships to a protein than a single blast could.)
Looks like this is a terminase, small subunit to me.
debbie

Link to this post \| posted 19 Feb, 2023 13:21
kmaclea	That is very helpful, Debbie! Thank you! My inclination was to ignore this particular NCBI RefSeq call–so I am glad to hear you agree. And in the meantime I am definitely hoping this isn't more widespread! Kyle – Kyle MacLea Associate Professor, University of New Hampshire at Manchester kyle.maclea@unh.edu +1 603-641-4129

Link to this post \| posted 22 Feb, 2023 20:00
aajohnson	debbie Hi Kyle, So if you continue to investigate, there is no supporting data for a minor tail protein except NCBI said so. there is no way that a functional call can be made from the blast data that HHPred data sources does not support." We're getting no blastp CDD or HHPred support for at least two "minor tail proteins" in cluster K phages, including in the publication supplementary tables posted there. I would call them as hypothetical protein. I haven't dug yet for any synteny rules or something related to these- can you point me towards something to use? I commiserate with you Kyle.

Link to this post | posted 22 Feb, 2023 20:00

aajohnson

debbie
Hi Kyle,
So if you continue to investigate, there is no supporting data for a minor tail protein except NCBI said so. there is no way that a functional call can be made from the blast data that HHPred data sources does not support."

We're getting no blastp CDD or HHPred support for at least two "minor tail proteins" in cluster K phages, including in the publication supplementary tables posted there. I would call them as hypothetical protein. I haven't dug yet for any synteny rules or something related to these- can you point me towards something to use?
I commiserate with you Kyle.

Link to this post \| posted 23 Feb, 2023 01:14
debbie	Hi Allison, Minor tail proteins are the most common functional assignments that are acceptable to call with no CDD or HHPred supporting data. The genes that are "eligible" are the 4-6 large genes downstream of tape measure. I would not make the assignment if the gene is not a relatively big gene. In the case of Kyle's question, there is discrepant data that contradicts a minor tail protein call - synteny and HHPred hits. Best, debbie

Link to this post | posted 23 Feb, 2023 01:14

debbie

Hi Allison,
Minor tail proteins are the most common functional assignments that are acceptable to call with no CDD or HHPred supporting data. The genes that are "eligible" are the 4-6 large genes downstream of tape measure. I would not make the assignment if the gene is not a relatively big gene.
In the case of Kyle's question, there is discrepant data that contradicts a minor tail protein call - synteny and HHPred hits.
Best,
debbie

Link to this post \| posted 23 Feb, 2023 13:40
aajohnson	Perfect, thank you.

Link to this post \| posted 23 Feb, 2023 17:33
kmaclea	Thank you–I like to use the MTPs as an example and it's nice to be reminded of the parameters around the function call! (This silly RefSeq issue notwithstanding!) – Kyle MacLea Associate Professor, University of New Hampshire at Manchester kyle.maclea@unh.edu +1 603-641-4129

Link to this post \| posted 24 Feb, 2023 20:52
kmaclea	So we've got another instance of the same issue, Debbie. MulchSalad_Draft_37 is an integrase/tyrosine integrase by analysis on Phagesdb BLAST/HHPred as shown in the pham report: https://capture.dropbox.com/TobVF9m4QnDFz6Hl NCBI BLAST is showing mostly "endonuclease" as the annotation: https://capture.dropbox.com/g0O9azXO34T0rhG9 And if you do the "Identical Proteins" analysis on any of the predicted "endonuclease" genes you see the difference again: https://capture.dropbox.com/RTPOWS6fziZ4TMpK What we're seeing again is the RefSeq ("curated" database is calling most of our Cluster F pham 68632 integrases as endonucleases whereas GenBank/INSDC and Phagesdb are calling them integrases. So somewhere along the way RefSeq "curation" is changing the default of what we are calling these genes. And since apparently, by default, NCBI BLAST shows the RefSeq results and only shows the GenBank results if you dig deeper, I'm guessing this will cause a certain amount of confusion going forward. Meh. Kyle – Kyle MacLea Associate Professor, University of New Hampshire at Manchester kyle.maclea@unh.edu +1 603-641-4129 Edited 24 Feb, 2023 20:54

Link to this post | posted 24 Feb, 2023 20:52

kmaclea

So we've got another instance of the same issue, Debbie.

MulchSalad_Draft_37 is an integrase/tyrosine integrase by analysis on Phagesdb BLAST/HHPred as shown in the pham report:

https://capture.dropbox.com/TobVF9m4QnDFz6Hl

NCBI BLAST is showing mostly "endonuclease" as the annotation:

https://capture.dropbox.com/g0O9azXO34T0rhG9

And if you do the "Identical Proteins" analysis on any of the predicted "endonuclease" genes you see the difference again:

https://capture.dropbox.com/RTPOWS6fziZ4TMpK

What we're seeing again is the RefSeq ("curated" smile

database is calling most of our Cluster F pham 68632 integrases as endonucleases whereas GenBank/INSDC and Phagesdb are calling them integrases. So somewhere along the way RefSeq "curation" is changing the default of what we are calling these genes.

And since apparently, by default, NCBI BLAST shows the RefSeq results and only shows the GenBank results if you dig deeper, I'm guessing this will cause a certain amount of confusion going forward.

Meh.

Kyle

–
Kyle MacLea
Associate Professor, University of New Hampshire at Manchester
kyle.maclea@unh.edu +1 603-641-4129

Edited 24 Feb, 2023 20:54

Link to this post \| posted 03 Mar, 2023 19:29
kmaclea	Just a note that we continue to find this in working on MulchSalad (F1) genes. We saw some genes today in which RefSeq is calling the gene one thing and INSDC is instead calling it a hypothetical protein! So it even gets more complicated! Kyle – Kyle MacLea Associate Professor, University of New Hampshire at Manchester kyle.maclea@unh.edu +1 603-641-4129

Recent Activity

RefSeq and INSDC name disagreements in NCBI Blast for Functonal Assignment