SEA-PHAGES | helix-turn-helix binding domain or protein?

Link to this post \| posted 04 Feb, 2022 19:19
uOttawaPHAGE	We are annotating a gene with a HTH DNA binding domain. However in the notes column of the approved list it refers to the function HTH DNA binding protein, even though only HTH DNA binding domain is on the approved function list. I'm pretty sure you consider these two things to be the same, but it was confusing to my students. Do you consider domain and protein different annotations?

Link to this post \| posted 04 Feb, 2022 20:21
debbie	I think I corrected the confusion on the approved function list. Note that 'protein' was never mentioned in the first column. The biology confusion is that "is a 'helix-turn-helix' a protein function?". That is the original of the use of domain. We are saying that this protein has a helix-turn-helix domain in it, but what it does (other than fit into DNA nicely), is varied. Does this help? debbie Edited 15 Feb, 2022 13:23

Link to this post | posted 04 Feb, 2022 20:21

debbie

I think I corrected the confusion on the approved function list. Note that 'protein' was never mentioned in the first column.
The biology confusion is that "is a 'helix-turn-helix' a protein function?". That is the original of the use of domain. We are saying that this protein has a helix-turn-helix domain in it, but what it does (other than fit into DNA nicely), is varied.
Does this help?
debbie

Edited 15 Feb, 2022 13:23

Link to this post \| posted 15 Feb, 2022 03:39
uOttawaPHAGE	Yes, I assumed you meant HTH DNA binding domain, not protein. It is just a motif, not a function. Similar, but unrelated question (probably belongs in another thread?): we have annotated several membrane proteins. Again a motif, not a function. However, it seems like membrane protein doesn't get added to phamerator maps of the comparator phages even though it is is the genbank file. Why is that?

Link to this post \| posted 15 Feb, 2022 13:38
debbie	Adam, At the crux of the membrane protein designation is "what constitutes a membrane protein?". We have established a rule. (Know that phages don't have rules, so this is problematic.) The rule is if a protein has 2 or more transmembrane domains, call it a membrane protein. However, if it only has one, you must confirm the transmembrane domain with another transmembrane finder. Most recommended finders are TMHMM and SOSUI. PECAAN is using TopCons, which might be more helpful if we could agree on how to interpret it. However, Phamerator does not remove anything protein labels, ever. What can happen is that at Pitt, I have a team who are curating the functional calls (discrepancies) by phams. Our 'corrections' go directly to the GenBank file. Phamerator systematically checks changes in GenBank files and updates. You are right, membrane protein is not a functional call. It can be informative. Unfortunately, over time we have waffled on the one transmembrane call so those calls may or may not be there.

Link to this post | posted 15 Feb, 2022 13:38

debbie

Adam,
At the crux of the membrane protein designation is "what constitutes a membrane protein?". We have established a rule. (Know that phages don't have rules, so this is problematic.) The rule is if a protein has 2 or more transmembrane domains, call it a membrane protein. However, if it only has one, you must confirm the transmembrane domain with another transmembrane finder. Most recommended finders are TMHMM and SOSUI. PECAAN is using TopCons, which might be more helpful if we could agree on how to interpret it.

However, Phamerator does not remove anything protein labels, ever. What can happen is that at Pitt, I have a team who are curating the functional calls (discrepancies) by phams. Our 'corrections' go directly to the GenBank file. Phamerator systematically checks changes in GenBank files and updates.

You are right, membrane protein is not a functional call. It can be informative. Unfortunately, over time we have waffled on the one transmembrane call so those calls may or may not be there.

Link to this post \| posted 11 Jan, 2023 19:45
jdaft@leeuniversity.edu	I was just curious as to what additional evidence I need to call a protein a "helix-turn-helix DNA binding domain, MerR-like" vs just a "helix-turn-helix DNA binding domain"? I'm including a link to my HHpred hit https://toolkit.tuebingen.mpg.de/jobs/9384933, it shows up as the second hit. NCBI blast just shows up as a helix-turn-helix dna binding domain. Is there some other tool I can use to make the call between the two? I know HTH dna binding domain is the safe call.

Link to this post | posted 11 Jan, 2023 19:45

jdaft@leeuniversity.edu

I was just curious as to what additional evidence I need to call a protein a "helix-turn-helix DNA binding domain, MerR-like" vs just a "helix-turn-helix DNA binding domain"? I'm including a link to my HHpred hit https://toolkit.tuebingen.mpg.de/jobs/9384933, it shows up as the second hit. NCBI blast just shows up as a helix-turn-helix dna binding domain. Is there some other tool I can use to make the call between the two? I know HTH dna binding domain is the safe call.

Link to this post \| posted 11 Jan, 2023 20:24
cdshaffer	when presented with this issue of specificity (i.e. should one annotate using the more specific term that includes the "MerR like" ) I like to first check out the PFam hits; others prefer the CD hit database, both are fine, I am just used to the Pfam dataset and I think the Hidden markov models that underlie PFam database (instead of the PSSM data in CDD) are more sensitive. You can add the Pfam database to an hhpred search, I usually don't add this database by default but I would go back and add it in a situation like this. In this case, your protein hits PF13411.9 ; MerR_1 ; MerR HTH family regulatory protein with 98% and 75% alignment. So in terms of assessing the quality of the hit I would say it is borderline (for PFam hits I do like nearly full length) but since this is just a specificity issue (I.e. there is plenty of evidence from your search that there is indeed an HTH in this protein) I would tell a student that, yes, this PFam hit is sufficient evidence to support the added "MerR like" term to the function. Note that some of the other proteins in your search do also mention the MerR like domain but you need to be careful and not take that as evidence at face value. Since HHPRED will happily report partial alignments, to use that evidence you would need to confirm that the MerR like HTH domain in the description is also in the region of the alignment. This can easily be done, it just usually takes a few clicks to dig into other databases and thus more time; whereas when looking for a good PFam hit, all the evidence you need is right there on the HHpred results page, so it is more of an expediency issue than a "quality of evidence" issue as to why I prefer using PFam. Edited 11 Jan, 2023 20:24

Link to this post | posted 11 Jan, 2023 20:24

cdshaffer

when presented with this issue of specificity (i.e. should one annotate using the more specific term that includes the "MerR like" ) I like to first check out the PFam hits; others prefer the CD hit database, both are fine, I am just used to the Pfam dataset and I think the Hidden markov models that underlie PFam database (instead of the PSSM data in CDD) are more sensitive.

You can add the Pfam database to an hhpred search, I usually don't add this database by default but I would go back and add it in a situation like this. In this case, your protein hits PF13411.9 ; MerR_1 ; MerR HTH family regulatory protein with 98% and 75% alignment. So in terms of assessing the quality of the hit I would say it is borderline (for PFam hits I do like nearly full length) but since this is just a specificity issue (I.e. there is plenty of evidence from your search that there is indeed an HTH in this protein) I would tell a student that, yes, this PFam hit is sufficient evidence to support the added "MerR like" term to the function.

Note that some of the other proteins in your search do also mention the MerR like domain but you need to be careful and not take that as evidence at face value. Since HHPRED will happily report partial alignments, to use that evidence you would need to confirm that the MerR like HTH domain in the description is also in the region of the alignment. This can easily be done, it just usually takes a few clicks to dig into other databases and thus more time; whereas when looking for a good PFam hit, all the evidence you need is right there on the HHpred results page, so it is more of an expediency issue than a "quality of evidence" issue as to why I prefer using PFam.

Edited 11 Jan, 2023 20:24

Link to this post \| posted 13 Jan, 2023 04:15
debbie	In our investigations, we determined that a MerR-like HTH will have 3 helix-coil motifs followed by a "wing' of a coil-coil. As I review your seqeunce int he link above, I do not see the 'wing' with a coil-coil at the end. so i would call this a helix-turn-helix DNA binding protein.

Link to this post \| posted 11 Apr, 2023 18:34
ACMPhageHunters	Hello, I'm having trouble with two gene function calls regarding HTH domains. In the first gene, many of the Phagesdb Blast hits call a "helix-turn-helix DNA binding domain protein" and exact matches on the NCBI results are "HTH DNA binding protein", but I don't see any references to an HTH binding domain in the HHpred or CDD results. The second gene has similar results. I don't want to just go with the crowd on these, so without HHpred or CDD confirmation, I'm leaning towards NKF. Is there anything else I should do to check for these HTH domains? The genes in question: StolenFromERC Cluster: E PECAAN # 32 Start 29,140 Stop: 28955 Reverse PECANN #130 Start: 68,963 Stop: 68700 Reverse Thanks for your help, Michèle

Link to this post | posted 11 Apr, 2023 18:34

ACMPhageHunters

Hello,
I'm having trouble with two gene function calls regarding HTH domains. In the first gene, many of the Phagesdb Blast hits call a "helix-turn-helix DNA binding domain protein" and exact matches on the NCBI results are "HTH DNA binding protein", but I don't see any references to an HTH binding domain in the HHpred or CDD results. The second gene has similar results. I don't want to just go with the crowd on these, so without HHpred or CDD confirmation, I'm leaning towards NKF. Is there anything else I should do to check for these HTH domains? The genes in question:
StolenFromERC Cluster: E
PECAAN # 32 Start 29,140 Stop: 28955 Reverse
PECANN #130 Start: 68,963 Stop: 68700 Reverse

Thanks for your help,
Michèle

Link to this post \| posted 11 Apr, 2023 19:17
uOttawaPHAGE	Hi Michèle. I quickly looked at the HHpred results for these two genes. IF you look down your list their are hits with HTH domains, and many of your hits are either sigma factors or related to sigma factors, which have HTH domains. Upon closer inspection the matches all appear to be minimal domains that have three (in one or two cases only two) alpha helical domains with short non-helical segments. I believe they both meet the threshold of having a HTH domain. I still struggle with what this function should be. I think "HTH domain" or ""HTH domain containing protein" would be best, but I don't thinks these are options on the approved function list. Many of the hits are to transcription factors that have HTH domains, so they could be "HTH DNA binding domain protein" but it does seem like there isn't sufficient data to call these "DNA binding proteins." Debbie will need to weigh in on this. It would seem like we need a new more minimal function for these types of proteins. Adam

Link to this post | posted 11 Apr, 2023 19:17

uOttawaPHAGE

Hi Michèle. I quickly looked at the HHpred results for these two genes. IF you look down your list their are hits with HTH domains, and many of your hits are either sigma factors or related to sigma factors, which have HTH domains. Upon closer inspection the matches all appear to be minimal domains that have three (in one or two cases only two) alpha helical domains with short non-helical segments. I believe they both meet the threshold of having a HTH domain.

I still struggle with what this function should be. I think "HTH domain" or ""HTH domain containing protein" would be best, but I don't thinks these are options on the approved function list. Many of the hits are to transcription factors that have HTH domains, so they could be "HTH DNA binding domain protein" but it does seem like there isn't sufficient data to call these "DNA binding proteins."

Debbie will need to weigh in on this. It would seem like we need a new more minimal function for these types of proteins.

Adam

Link to this post \| posted 12 Apr, 2023 17:18
cdshaffer	#32 has a clear and strong HHPRED match to the several different HTH pfams and as Adam mentioned HHPREd does predict helix-turn-helix like secondary structure for the amino acids around 19 - 55. so i would add an annotation of "helix-turn-helix DNA binding domain". for me, I teach my students that these words we are adding are "annotations" not necessarily "functions". Ideally (but only rarely) is the evidence sufficient to be convinced what the exact biological role is for a protein. But that does not mean that we cannot add value by adding annotations that help the readers of our "publication" (i.e. genbank entries). So adding "helix-turn-helix DNA binding domain" annotation helps the reader limit the possible roles (and is therefore a good annotation) even if the evidence is not sufficient to generate a better more informative annotation like "sigma factor" etc. If a protein has good evidence for an HTH domain I assume there is a high probability it does indeed bind DNA but I typically only use the approved term "DNA binding protein" if I find good evidence for a DNA binding motif that is not an HTH like a zinc finger, leucine zipper or talons. If I see an HTH domain, I prefer the "helix-turn-helix DNA binding domain" over "DNA binding" even though it very likely does bind DNA if it has a HTH, I will just cite the evidence (matches HTH domain) and leave the assumption of activity (actually binds DNA) up to the reader.

Link to this post | posted 12 Apr, 2023 17:18

cdshaffer

#32 has a clear and strong HHPRED match to the several different HTH pfams and as Adam mentioned HHPREd does predict helix-turn-helix like secondary structure for the amino acids around 19 - 55. so i would add an annotation of "helix-turn-helix DNA binding domain".

for me, I teach my students that these words we are adding are "annotations" not necessarily "functions". Ideally (but only rarely) is the evidence sufficient to be convinced what the exact biological role is for a protein. But that does not mean that we cannot add value by adding annotations that help the readers of our "publication" (i.e. genbank entries). So adding "helix-turn-helix DNA binding domain" annotation helps the reader limit the possible roles (and is therefore a good annotation) even if the evidence is not sufficient to generate a better more informative annotation like "sigma factor" etc.

If a protein has good evidence for an HTH domain I assume there is a high probability it does indeed bind DNA but I typically only use the approved term "DNA binding protein" if I find good evidence for a DNA binding motif that is not an HTH like a zinc finger, leucine zipper or talons. If I see an HTH domain, I prefer the "helix-turn-helix DNA binding domain" over "DNA binding" even though it very likely does bind DNA if it has a HTH, I will just cite the evidence (matches HTH domain) and leave the assumption of activity (actually binds DNA) up to the reader.

Recent Activity

helix-turn-helix binding domain or protein?