A bug with missing right-side type bulges

[xiangjun]

This thread is to replace the original one titled "an issue with bulges" stared by febos. Somehow, the URL with the original thread does not reflect the topic, and trying to fix the issue leads to a broken link when clicking directly on the topic subject. [Note added on 2014-01-10: the broken URL bug has been fixed.]

The original thread with three posts is still kept for record, and this new thread is a direct copy of the original posts.

Xiang-Jun

[xiangjun]

Good Afternoon, Dr. Lu.

I think I've found a real bug with bulges in DSSR:

Bulges can be of two types - "right" ([0xN]) and "left" ([Nx0]).  I looked at all pdb files with RNA and found that in DSSR output there are just 4 cases of right bulges (from about 8500 bulges) and all these 4 cases are bulges between different rna chains! (strand1 - one chain; strand2 - another chain). I'm sorry but now I can't remember in which files these 4 cases were. But the main problem is that DSSR has no proper right bulges!

One example:

file 1FFK, chain 9:

I attached picture of its secondary structure (1FFK_9.png, made with VARNA).

As you can see from picture there are 3 bulges on this structure:

1) .9.A.65.                (right bulge of form [0x1])
2) .9.A.51.+.9.A.52. (right bulge of form [0x2])
3) .9.U.87.                (left bulge of form  [1x0])

I also attached DSSR ouput (1FFK.out1), from which you can see:

bulge number 3 ([1x0]):

Code: [Select]

  12 bulge: 5 nts; [1x0]; linked by [#174, #175]
       .9.G.86.+.9.U.87.+.9.G.88.+.9.C.95.+.9.C.96. [GUGCC]
       1 nts bulge .9.U.87. [U]; .9.G.86.-->.9.G.88. [GUG]
       0 nts bulge ; .9.C.95.-->.9.C.96. [CC]
so far so good
and now bulges number 1 and 2 are in non-loop single-stranded segments section:

Code: [Select]

  93 nts=2 .9.A.51.+.9.A.52. [AA]
  94 nts=1 .9.A.65. [A]
I know that 1FFK is not the only file where right bulges exist, so I think that's a common bug.
I hope my explanation was useful.

Best regards,
Eugene
Originally posted on: December 24, 2013, 05:57:21 AM

[xiangjun]

Hi Eugene,

Thanks for reporting the bulging issue. Your example (PDB id 1ffk, chain 9) clearly illustrates a bug in DSSR for detecting bulges. The nts missed in right bulges are classified as single-stranded. So the two manifestations are of the same origin.

Over the past couple of weeks, I've actually been thinking about tidying up the algorithm for identifying the various loops and presenting them consistently. Your bug report fits just in time. I will try to fix the issue soon.

Best regards,

Xiang-Jun
Originally posted on December 25, 2013, 11:30:20 AM

[xiangjun]

Hi Eugene,

The missing right-type bulges bug has been fixed in DSSR beta-r28-on-20131225. Please have a try and report back how it goes.

For PDB entry 1ffk, chain 9, the corresponding output section now looks like below:

Code: [Select]

****************************************************************************
List of 2 hairpin loop(s)
   1 nts=4 GCGA closed by pair {9.C89+9.G94 [CG], #8}
       9.C89+9.G90+9.C91+9.G92+9.A93+9.G94 [CGCGAG]
   2 nts=5 AUCCC closed by pair {9.C37+9.G43 [CG], #-1}
       9.C37+9.A38+9.U39+9.C40+9.C41+9.C42+9.G43 [CAUCCCG]

****************************************************************************
List of 3 bulge(s)
   1 bulge: 5 nts; [0x1]; linked by [#2, #3]
       9.C15+9.G16+9.C64+9.A65+9.G66 [CGCAG]
       0 nts bulge ; 9.C15-->9.G16 [CG]
       1 nts bulge 9.A65 [A]; 9.C64-->9.G66 [CAG]
   2 bulge: 6 nts; [0x2]; linked by [#4, #5]
       9.C29+9.C30+9.G50+9.A51+9.A52+9.G53 [CCGAAG]
       0 nts bulge ; 9.C29-->9.C30 [CC]
       2 nts bulge 9.A51+9.A52 [AA]; 9.G50-->9.G53 [GAAG]
   3 bulge: 5 nts; [1x0]; linked by [#7, #8]
       9.G86+9.U87+9.G88+9.C95+9.C96 [GUGCC]
       1 nts bulge 9.U87 [U]; 9.G86-->9.G88 [GUG]
       0 nts bulge ; 9.C95-->9.C96 [CC]

****************************************************************************
List of 3 internal loop(s)
   1 asymmetric internal loop: 14 nts; [6x4]; linked by [#3, #4]
       9.G21+9.G22+9.U23+9.U24+9.G25+9.C26+9.C27+9.U28+9.A54+9.U55+9.A56+9.A57+9.G58+9.C59 [GGUUGCCUAUAAGC]
       6 nts iloop 9.G22+9.U23+9.U24+9.G25+9.C26+9.C27 [GUUGCC]; 9.G21-->9.U28 [GGUUGCCU]
       4 nts iloop 9.U55+9.A56+9.A57+9.G58 [UAAG]; 9.A54-->9.C59 [AUAAGC]
   2 symmetric internal loop: 10 nts; [3x3]; linked by [#5, #-1]
       9.U33+9.A34+9.C35+9.C36+9.C37+9.G43+9.A44+9.A45+9.C46+9.A47 [UACCCGAACA]
       3 nts iloop 9.A34+9.C35+9.C36 [ACC]; 9.U33-->9.C37 [UACCC]
       3 nts iloop 9.A44+9.A45+9.C46 [AAC]; 9.G43-->9.A47 [GAACA]
   3 asymmetric internal loop: 13 nts; [5x4]; linked by [#6, #7]
       9.G75+9.G76+9.A77+9.G78+9.U79+9.A80+9.C81+9.G101+9.G102+9.A103+9.A104+9.A105+9.C106 [GGAGUACGGAAAC]
       5 nts iloop 9.G76+9.A77+9.G78+9.U79+9.A80 [GAGUA]; 9.G75-->9.C81 [GGAGUAC]
       4 nts iloop 9.G102+9.A103+9.A104+9.A105 [GAAA]; 9.G101-->9.C106 [GGAAAC]

****************************************************************************
List of 1 junction(s)
   1 3-way junctions: 16 nts; [4x3x3]; linked by [#1, #2, #6]
       9.C9+9.C10+9.A11+9.C12+9.A13+9.G14+9.C67+9.G68+9.U69+9.U70+9.C71+9.G110+9.U111+9.U112+9.C113+9.G114 [CCACAGCGUUCGUUCG]
       4 nts junction 9.C10+9.A11+9.C12+9.A13 [CACA]; 9.C9-->9.G14 [CCACAG]
       3 nts junction 9.G68+9.U69+9.U70 [GUU]; 9.C67-->9.C71 [CGUUC]
       3 nts junction 9.U111+9.U112+9.C113 [UUC]; 9.G110-->9.G114 [GUUCG]

****************************************************************************
List of 2 non-loop single-stranded segment(s)
   1 nts=3 9.U1-->9.A3 [UUA]
   2 nts=3 9.A120-->9.C122 [ACC]

Again, thanks for reporting the bug!

Xiang-Jun
Originally posted on: December 25, 2013, 03:00:25 PM

[febos]

Hi Dr. Lu

I'm glad to help you.

Now it works fine.

Best regards,
Eugene.