infinity enegies

1 reply [Last post]
revyakin
Offline
Joined: 04/11/2013

Hi all again,

Meant to ask this for a while. With some sequences, I get infinity energies in my analyses. I currently weed them out at the parsing stage, but I am wondering what the reason for the infinity data is, to make sure my analyses is not biased. Sequences are shown below (sometimes up to 20% of energies are infinities)

>hybrid-ss-min -E --NA=DNA -M=0.01 -N=0.15

#T -RT ln Z Z
37 2.57 0.0154547
37 2.95 0.00834247
37 3.67 0.00259388
37 3.27 0.00496373
37 999.999 0 AACTCCAACCTCCCCATCT;
37 3.15 0.00603067
37 2.57 0.0154547
37 999.999 0 CCTCTTCACCCACACTATT
37 999.999 0 AAACCCAAACCATCCTCAA
...

I wonder if it has anything to do with the -M=0.01 -N=0.15 parameters, because the log file shrank these to zero (I meant to run at 10 mM Mg and 150 mM NaCl):

hybrid-ss-min 3.8 ran on .... at Mon Sep 30 16:32:38 2013

NA = DNA
tMin = 37
tInc = 1
tMax = 37
[Na+] = 0
[Mg++] = 0
maxloop = 30
prefilter 2/2

This is the sequence file (top 9 lines)
CTACCCCCTATCTCATACA;
CCATTAACCCAACCCACTA;
CCATCCCTTCCCTATATCA;
TCCAATCCCCATTCTCCCC;
AACTCCAACCTCCCCATCT;
CCCAATTATCCAACCCACT;
CCCATCTCAATTCTATACC;
CCTCTTCACCCACACTATT;
AAACCCAAACCATCCTCAA;
...

Running problemtaic sequences with default parameters produced "infinity" numbers, but with decimal points changed:
#T -RT ln Z Z
37 999.349 0
37 999.149 0
37 999.349 0

NMarkham
Offline
Joined: 09/29/2011
I'm committing something of a

I'm committing something of a faux pas in replying to a post that's so old, but I figured better late than never. :)

  1. In general, an energy of +infinity (or any large positive number) or a partition function of zero means no structure is possible. This sometimes happens unexpectedly because UNAFold requires (by default) at least two base pairs in a row for a structure. A sequence like AACTCCAACCTCCCCATCT has no possible G-C (or G-T) pairs and only a handful of possible A-T pairs. Since none of those pairs are even potentially part of a 2+ base pair helix, UNAFold considers there to be no possible structure. (Even if it did consider a structure with one or two isolated base pairs, the free energy would be so unfavorable that the end result would be practically the same.)
  2. On an unrelated note, I should point out that -M=0.01 -N=0.15 is not valid syntax! UNAFold conforms to a fairly common standard among Linux/Unix software, where there are four ways to express an option with a value:
    • -M0.01
    • -M 0.01
    • --Magnesium=0.01
    • --Magnesium 0.01

    An expression like -M=0.01 is interpreted as an attempt to set [Mg++] to "=0.01", which is nonsense. I regret that UNAFold silently turns such a string into zero, rather than printing an error message.