Stems of junction structure have only one base pair

[lijun]

Hi all,

I am using the newest DSSR to extract the junction structure, but the stems of some junctions have only one base pair. Take 5vyc.cif for an example and one of its 3-way junctions is as follows:

   1 3-way junction: nts=11; [2,1,2]; linked by [#63,#64,#-49]
     summary: [3] 2 1 2 [i2.1225 i2.1641 i2.1228 i2.1530 i2.1532 i2.1638] 5 3 1
     nts=11 UGGAUACGGAA 1..i2.U.1225.,1..i2.G.1226.,1..i2.G.1227.,1..i2.A.1228.,1..i2.U.1530.,1..i2.A.1531.,1..i2.C.1532.,1..i2.G.1638.,1..i2.G.1639.,1..i2.A.1640.,1..i2.A.1641.
       nts=2 GG 1..i2.G.1226.,1..i2.G.1227.
       nts=1 A 1..i2.A.1531.
       nts=2 GA 1..i2.G.1639.,1..i2.A.1640.

The last stem #-49 has only one base pair. There are many cases like this in the output. Is it reasonable to take structure like this as a junction?

Thanks,

Jun Li

[xiangjun]

Hi Jun Li,

Thanks for using DSSR and for posting your questions on the 3DNA Forum.

I understand that the inclusion of isolated canonical pairs (icPairs, those with negative index numbers) in delineating loops may lead to 'spurious' (junction) loops. I am so glad that this issue has been finally brought up on the Forum.

As far as I know, no consensus exists as to whether icPairs should be (always) included in the delineation of loops. Without taking them into consideration, however, the widely documented GUAA tetraloop in PDB entry 1msy (see figure below) cannot be logically identified. That's why DSSR counts icPairs in loops, by default.

While undocumented, DSSR has an option to exclude icPairs from loops, i.e. (--loop-exclude-icpair). Run DSSR with this option on your example PDB entry 5vyc, you'd find that the 3-way junction you reported is gone.

In a similar way, it is also debatable if a non-canonical base pair in an otherwise 'regular' double helix should be taken as part of a stem or as an internal loop. As noted in the 2015 DSSR paper in NAR:

Following standard conventions (13,35–36), DSSR by default uses only canonical pairs to identify stems and to characterize pseudoknots. Nevertheless, the program contains provisions for including noncanonical pairs in an extended definition of stems

It is my hope that DSSR would help 'standardize' the field, probably in the long run.

Best regards,

Xiang-Jun

[lijun]

Hi Xiang-Jun,

Thanks for your prompt reply and help.

Jun Li

Hi Jun Li,

Thanks for using DSSR and for posting your questions on the 3DNA Forum.

I understand that the inclusion of isolated canonical pairs (icPairs, those with negative index numbers) in delineating loops may lead to 'spurious' (junction) loops. I am so glad that this issue has been finally brought up on the Forum.

As far as I know, no consensus exists as to whether icPairs should be (always) included in the delineation of loops. Without taking them into consideration, however, the widely documented GUAA tetraloop in PDB entry 1msy (see figure below) cannot be logically identified. That's why DSSR counts icPairs in loops, by default.

While undocumented, DSSR has an option to exclude icPairs from loops, i.e. (--loop-exclude-icpair). Run DSSR with this option on your example PDB entry 5vyc, you'd find that the 3-way junction you reported is gone.

In a similar way, it is also debatable if a non-canonical base pair in an otherwise 'regular' double helix should be taken as part of a stem or as an internal loop. As noted in the 2015 DSSR paper in NAR:

Following standard conventions (13,35–36), DSSR by default uses only canonical pairs to identify stems and to characterize pseudoknots. Nevertheless, the program contains provisions for including noncanonical pairs in an extended definition of stems

It is my hope that DSSR would help 'standardize' the field, probably in the long run.

Best regards,

Xiang-Jun

[xiangjun]

You're welcome. As mentioned on the Forum and in the manual, DSSR contains many undocumented 'features'. User questions on subtle/technical aspects in DSSR are always gladly received.

As a side note, by including isPairs in loops by default, it is possible to filter DSSR output as users see fit. For example, by parsing for the negative index (#-49), or no. of pairs in a stem (1), users can easily remove the 3-way junction.

1 3-way junction: nts=11; [2,1,2]; linked by [#63,#64,#-49]
     summary: [3] 2 1 2 [i2.1225 i2.1641 i2.1228 i2.1530 i2.1532 i2.1638] 5 3 1

Along the line, it is also feasible to select junction loops with a stem length of e.g. 3 or more canonical pairs.

Best regards,

Xiang-Jun