SEA-PHAGES | tail chaperones in cluster BH

Link to this post \| posted 04 Aug, 2018 04:42
ivanerill	We suspect that the canonical arrangement of tail chaperones preceding the tapemeasure gene may be conserved in cluster BH, and possibly contain a programmed frameshift. We would like feedback on whether to annotate overlapping genes and/or help in detecting putative frameshift points. Please see attachment for details. 872Kb

Link to this post \| posted 05 Aug, 2018 00:47
lhughes	Hi Ivan, I guess I have a couple of questions on this, as I've tended to err on the side of not calling things unless I felt there was plenty of evidence to make the call. Synteny would say that these are in the correct location to be the tail assembly chaperone proteins. However, the first ORF has no hits at all to any other TAC proteins on either BLAST or HHPRED. To me, the other evidence you provide becomes less certain without this (though quite possibly correct). The inability to find the potential slippery sequence complicates this further. Another question I have relates to the potential DnaJ homology you discuss. Are DnaJ chaperones anything like the chaperone function of TAC proteins (ie. is this an actual relationship of function or not)? I guess part of my conundrum is whether it is "better" to call the long overlap or to call it shorter, given that both are probably wrong. I also don't know if we have enough evidence to make the function call at this point in time (but that is my tendency to call more conservatively). Lee

Link to this post | posted 05 Aug, 2018 00:47

lhughes

Hi Ivan,

I guess I have a couple of questions on this, as I've tended to err on the side of not calling things unless I felt there was plenty of evidence to make the call.

Synteny would say that these are in the correct location to be the tail assembly chaperone proteins. However, the first ORF has no hits at all to any other TAC proteins on either BLAST or HHPRED. To me, the other evidence you provide becomes less certain without this (though quite possibly correct). The inability to find the potential slippery sequence complicates this further.

Another question I have relates to the potential DnaJ homology you discuss. Are DnaJ chaperones anything like the chaperone function of TAC proteins (ie. is this an actual relationship of function or not)?

I guess part of my conundrum is whether it is "better" to call the long overlap or to call it shorter, given that both are probably wrong. I also don't know if we have enough evidence to make the function call at this point in time (but that is my tendency to call more conservatively).

Lee

Link to this post \| posted 06 Aug, 2018 14:59
ivanerill	Hi Lee, I agree that neither call would be completely correct. The main point of our argument is that it is hard to explain why a non-called region would have a good HHpred hit with a DnaJ chaperone domain (that is, why would a non-coding region match a protein domain). Given that synteny suggests these should be tail chaperones, the HHpred hit is harder to disregard. Hence my take would be to make the longer call. As you point out, this is likely to be wrong, since it seems logical to assume that if these two are indeed tail chaperones, one would strongly suspect that they would also be using translational frameshifting. I, however, could not find conclusive evidence of this. Regarding DnaJ, I am not aware of any formal link between them and TACs. I was able to find a reference that indicates that DnaJ is used by phages during tail assembly (http://www.jmb.or.kr/journal/download.php?Filedir=../submission/Journal/014/&num=764). My overall take on this is that these are tail chaperones (given their syntenic arrangement, coding potential evidence for an overlap and the presence of a conserved chaperone domain in the region that would be left uncalled using GeneMark), but that they are divergent enough to not get proper hits in HHpred (beyond the DnaJ chaperone hit in the non-called region). Ivan

Link to this post | posted 06 Aug, 2018 14:59

ivanerill

Hi Lee,

I agree that neither call would be completely correct. The main point of our argument is that it is hard to explain why a non-called region would have a good HHpred hit with a DnaJ chaperone domain (that is, why would a non-coding region match a protein domain). Given that synteny suggests these should be tail chaperones, the HHpred hit is harder to disregard. Hence my take would be to make the longer call. As you point out, this is likely to be wrong, since it seems logical to assume that if these two are indeed tail chaperones, one would strongly suspect that they would also be using translational frameshifting. I, however, could not find conclusive evidence of this.

Regarding DnaJ, I am not aware of any formal link between them and TACs. I was able to find a reference that indicates that DnaJ is used by phages during tail assembly (http://www.jmb.or.kr/journal/download.php?Filedir=../submission/Journal/014/&num=764).
My overall take on this is that these are tail chaperones (given their syntenic arrangement, coding potential evidence for an overlap and the presence of a conserved chaperone domain in the region that would be left uncalled using GeneMark), but that they are divergent enough to not get proper hits in HHpred (beyond the DnaJ chaperone hit in the non-called region).

Ivan

Link to this post \| posted 06 Aug, 2018 15:48
delesall	Hi Ivan Nice detective work on this. My major criterion in situations like this is whether if you make the wrong call, it will be easily detected or propagated through the database without detection. Because of that, I would not call function as I have QC.ed too many annotations where folks just see a function in blastp data and assume it is correct without further checking. However, a wrong start should be more easy to detect and although rare longer overlap between genes can be OK. So, I would probably go with the longer call that includes all the CP. Does starterator suggest that this start is conserved in other BH phages? And it is gene 23 that you want to make longer, right? Veronique

Link to this post | posted 06 Aug, 2018 15:48

delesall

Hi Ivan

Nice detective work on this. My major criterion in situations like this is whether if you make the wrong call, it will be easily detected or propagated through the database without detection. Because of that, I would not call function as I have QC.ed too many annotations where folks just see a function in blastp data and assume it is correct without further checking. However, a wrong start should be more easy to detect and although rare longer overlap between genes can be OK. So, I would probably go with the longer call that includes all the CP. Does starterator suggest that this start is conserved in other BH phages? And it is gene 23 that you want to make longer, right?

Veronique

Link to this post \| posted 06 Aug, 2018 18:38
ivanerill	Hi Veronique, Thanks. Yes, our idea was not so much to make the functional annotation (we agree the evidence is relatively, weak although the HHpred hit for the non-called region is over the p=0.9 threshold and synteny points toward these being tail chaperones), but to make the longer gene call. In all BH phages the option to make the longer call on Gp23 homologs is available, and results in a similar overlap with the previous gene (Gp22). Ivan

Link to this post | posted 06 Aug, 2018 18:38

ivanerill

Hi Veronique,

Thanks. Yes, our idea was not so much to make the functional annotation (we agree the evidence is relatively, weak although the HHpred hit for the non-called region is over the p=0.9 threshold and synteny points toward these being tail chaperones), but to make the longer gene call. In all BH phages the option to make the longer call on Gp23 homologs is available, and results in a similar overlap with the previous gene (Gp22).

Ivan

Link to this post \| posted 06 Aug, 2018 19:14
lhughes	I would definitely be on-board with the idea of making the longer call for the second gene (even with the overlap) much like we do with the two overlapping DNA primase genes in many other phage clusters. In this case as in that one, we believe there is a frameshift that we just can't identify, so the longer call would capture more of the information. I think we would be better off not calling the function at this time until there is more evidence to support the call. Debbie/Welkin - if we want to make a change across the board for the genomes already in GenBank/Phamerated, what is your recommended process? Lee

Link to this post | posted 06 Aug, 2018 19:14

lhughes

I would definitely be on-board with the idea of making the longer call for the second gene (even with the overlap) much like we do with the two overlapping DNA primase genes in many other phage clusters. In this case as in that one, we believe there is a frameshift that we just can't identify, so the longer call would capture more of the information. I think we would be better off not calling the function at this time until there is more evidence to support the call.

Debbie/Welkin - if we want to make a change across the board for the genomes already in GenBank/Phamerated, what is your recommended process?

Lee

Link to this post \| posted 07 Aug, 2018 16:03
welkin	So this is a multi-part question: one, the logistics, and two, the timeline. logistically, changing a file means submitting a five-column feature table that contains all the features and locus tags of the current records, only with the edits you would like. once you have this file and have edited it, we can send it to GenBank and request an update to the record. A caveat to that would be for records that have not yet gone live; in that case, we could probably make a new asn1 file and ask to replace the older asn1 file with the new one. timeline wise—At the moment, we'd prefer that you let us submit all the changes to the files in GenBank– the staff has requested that we not try to change the same file too many times in a row (they'd prefer a single file with all updates if possible). As we are in the middle of trying to reconcile a lot of genomes right now it would probably not be great for you to send them something and then me to send them something else the next week. so if you would like to collect all the current files and adit them and make the five column tables, I can take a look, and submit the changes along with any others we've got for these. Welkin

Link to this post | posted 07 Aug, 2018 16:03

welkin

So this is a multi-part question: one, the logistics, and two, the timeline.

logistically, changing a file means submitting a five-column feature table that contains all the features and locus tags of the current records, only with the edits you would like. once you have this file and have edited it, we can send it to GenBank and request an update to the record.

A caveat to that would be for records that have not yet gone live; in that case, we could probably make a new asn1 file and ask to replace the older asn1 file with the new one.

timeline wise—At the moment, we'd prefer that you let us submit all the changes to the files in GenBank– the staff has requested that we not try to change the same file too many times in a row (they'd prefer a single file with all updates if possible).
As we are in the middle of trying to reconcile a lot of genomes right now it would probably not be great for you to send them something and then me to send them something else the next week.

so if you would like to collect all the current files and adit them and make the five column tables, I can take a look, and submit the changes along with any others we've got for these.

Welkin

Link to this post \| posted 07 Aug, 2018 16:42
ivanerill	OK. Microdon has not been yet submitted. There are I believe a couple more on the BH cluster that are still draft. The rest were just released by GenBank this July…

Link to this post \| posted 07 Aug, 2018 16:44
welkin	ok. Then I think for now, I will link this thread to the Cluster specific forum for BH– that should help with all future annotations. I am sorry I can't give you a clearer timeline for the ones that are currently in.

Recent Activity

tail chaperones in cluster BH