errata2.list

mfold

File USR$DISK:[ZUKER.MFOLD]CIRC.FOR;6
  438         if (force(i).eq.2.and.force(j).eq.2) nforce = nforce + 1
  439   c     Perhaps I,J stacks over I+1,J-1?
  440         if (e.eq.erg(2,i,j,i+1,j-1)
  441        . + v(i+1,j-1)) then
******
File USR$DISK:[ZUKER.MFOLD]CIRC.FOR;4
  437   c     Perhaps I,J stacks over I+1,J-1?
  438         if (e.eq.erg(2,i,j,i+1,j-1)
  439        . + v(i+1,j-1)) then
************
************
File USR$DISK:[ZUKER.MFOLD]CIRC.FOR;6
  455         if (force(i).eq.2.and.force(j).eq.2) ep = ep - eparam(9)
  456    
  457         k = i+2
  458   c     Perhaps I,J closes a multi-loop?
******
File USR$DISK:[ZUKER.MFOLD]CIRC.FOR;4
  453    
  454         k = i+2
  455   c     Perhaps I,J closes a multi-loop?
************

5. February 27, 1990.

Bug in EFN for circular foldings. The initial base-pair was being 
treated as a closing pair for a multi-loop and was given the 
eparam(10) penalty energy. The solution is to set 'open' to 1 in the 
calling statement.
************ new
File USR$DISK:[ZUKER.MFOLD]EFN.FOR;25
  151            call efn(nergy,basepr(i),n+i,1)
  152            energy = ( energy + float(nergy) )/10.0
******   old
File USR$DISK:[ZUKER.MFOLD]EFN.FOR;24
  151            call efn(nergy,basepr(i),n+i,0)
  152            energy = ( energy + float(nergy) )/10.0
************

6. June 7, 1990.

Small bug in batgen.f : attempt to write to standard input.

290,291c290,291  < old     > new
<         write (outunit,*) 'That save file doesn''t exist.'
<         write (outunit,*) 'Continue anyway?'
---
>         write (inunit,*) 'That save file doesn''t exist.'
>         write (inunit,*) 'Continue anyway?'

7. July 10, 1990.

Bug in mrna.f . Undefined dangling energies for single base stacking 
should be set to zero instead of infinity. This bug can cause 
traceback failures when modified bases are used (e.g lowercase : a, 
c, g, or u; or symbols such as X). The fixup is simple.

948c948  
<               dangle(a,b,c,d) = infinity
---
>               dangle(a,b,c,d) = 0
969c969
<                      dangle(var1,var2,var3,var4) = j
---
>                      if (j.ne.infinity) dangle(var1,var2,var3,var4) = j

8. July 12, 1990.

Bug in efn.f . Closed excisions were being given the bonus energy of
eparam(9) instead of a zero energy.

------old-----------------------------------------
400   size = j-i-1
      if (size.eq.3) then
c        Check for closed excision
         if(seq(i+1).eq.' '.and.seq(i+2).eq.' '.and.seq(i+3).eq.' ') then
            erg = eparam(9)
            return
--------new----------------------------------------
400   size = j-i-1
      if (size.eq.3) then
c        Check for closed excision
         if(seq(i+1).eq.' '.and.seq(i+2).eq.' '.and.seq(i+3).eq.' ') then
            erg = 0
            return
---------------------------------------------------

9. August 8, 1990.

Danielle Konings at the NIH noticed a small bug in the gettloops 
subroutine in mrna.f . Format statement number 1 should be
1      format(/)
instead of
1      format(//)
This bug meant that the first special tetraloop (GAAA) was being 
ignored.

10. March 1, 1991.

New version of IRIX objects to a parameter statement in prompt.f
The X-11 versions never contained this bug.
This should be replaced as follows:

14,17c14,17 ==> new version
<       parameter (CTRL$H = char(8),
<      $           CTRL$J = char(10),
<      $           CTRL$M = char(13),
<      $           CTRL$U = char(21) )
--- ===> old version
>       parameter (CTRL$H = char(ichar('H')-ichar('@')),
>      $           CTRL$J = char(ichar('J')-ichar('@')),
>      $           CTRL$M = char(ichar('M')-ichar('@')),
>      $           CTRL$U = char(ichar('U')-ichar('@'))  )

11. April 3, 1991.

IRIX 3.2.2 shifts the labelling of the energy dot plot. This is 
corrected by the addition of a single line in dotplot.f :
         write (st,fmt='(I5)') hstnum(j)
c   Zuker corrects small label bug for IRIX 3.2.2 - extra line added on April 3, 1991.
         vert1(1) = float(j-ct)
         call cmov2(vert1(1),vert1(2))          !move graphic character pointer to (x,y)

12. August 21, 1991.

Joanne Ridgeway rewrote formid.f, multid.f and getseq.f for opening, 
examining and reading sequence files. These three files are 
completely rewritten. This rewrite corrects many poor coding 
practices in the old versions, and enables the user to read sequences
in the GenBank, EMBL, PIR, Stanford or GCG formats. A few changes are
needed in mrna.f, as outlined below.

Difference file for : 

< old mrna.f
  versus
> new mrna.f
===============================================================

686,687d685
<       logical used        !excise these two lines
<       used = .false.

716c714,717
<          if (cntrl(7).eq.2) call mseq(cntrl(5))
---
>          if (cntrl(7).eq.2) then
>              cntrl(5) = 0
>              call mseq(cntrl(5))
>          endif

758c759
<             call formid(seqlab,seq,n,maxsiz,used)
---
>             call formid(seqlab,seq,n,maxsiz)

791,792d791
<       logical used             !excise these two lines
<       data used/.false./

794,795c793,794
<       if (.not.used) then
<          call multid(seqlab,seq,n,maxsiz,used,i)
---
>       if (i.eq.0) then
>          call multid(seqlab,seq,n,maxsiz,i)

798c797
<          call multid(seqlab,seq,n,maxsiz,used,i)
---
>          call multid(seqlab,seq,n,maxsiz,i)

13. October 23, 1992.

The line :
      vmin  = infinity
was accidentally commented out in the fill subroutine (in lin.f). 
This caused an error if the minimum folding energy of a fragment was 
positive.

14. December 4, 1992.

A tiny bug was found in the heap_sort subroutine. The last two i,j
base pairs selected automatically by the computer (automatic
traceback) were transposed during the sort. This bug meant that there
was a remote possibility that the second to last of the predicted
suboptimal structures would have a lower energy than the last
structure.

Bug in subroutine heap_sort in sort.f
87c87
<       do ir = num-1,1,-1      ! new code : continue to swap down to 1
---
>       do ir = num-1,2,-1      ! old code


15. January 8, 1993.

The pnumplot subroutine was found to produce total rubbish. The old 
VAX version is correct. Roland Gaboury ported this incorrectly. 

  < : old               > : new
643,646c643,648
<             if ((v(i,j) + v(j,i+n)).le.(vmin+vinc)) then
<                sum = sum + 1
<             end if
<          end do
---
>             if (j.lt.i) then
>                if ((v(j,i) + v(i,j+n)).le.(vmin+vinc)) sum = sum + 1
>             else
>                if ((v(i,j) + v(j,i+n)).le.(vmin+vinc)) sum = sum + 1
>             endif
>          enddo
675,678c677,683
<             if ((v(i,j) + v(j,i+n)).le.(vmin+vinc)) then
<                sum = sum + 1
<             end if
<          end do
---
>             if (j.lt.i) then
>                if ((v(j,i) + v(i,j+n)).le.(vmin+vinc)) sum = sum + 1
>             else
>                if ((v(i,j) + v(j,i+n)).le.(vmin+vinc)) sum = sum + 1
>             endif
>          enddo

16. Sep 27, 1995.

Finally cleaned up default names for output. Set 'dstr' to
'                                        ' in a data statement
in mrna.f and clean up the '.ct' assignment to '.ct '.
 

17. Jan 25, 1996.

Increased MAXREC from 80 to 120 in dot2ps.h
Long sequence names were causing a problem on the mfold server.