Change summary from MUSCLE version 3.8.31 to 3.8.31b

Several bug fixes and performance enhancements (e.g. speed, memory
usage) have been made to MUSCLE. Changes marked with a '*' are
considered more important. For example, pairwise distance calculations
have been accelerated with SSE instructions (when supported) for a
30-50x speedup. This speedup is significant when aligning data sets with
thousands of sequences (e.g. 10,000+ flu viruses). Unfortunately the
assembly is GCC specific. Another improvement packs the traceback
pointers of dynamic programming into nibbles, thus saving half the
memory. This is significant when aligning sequences with a length
greater than 10,000 (e.g. Herpes virus at 167,043). Also, using a "char"
type instead of "int" for m_uSortOrder in profile.h saves a significant
amount of memory.

Changes have been made to pave the way for dynamic allocation of the
struct ProfPos arrays (m_uSortOrder[], m_fcCounts[], m_AAScores[]) in
profile.h. All code will now only access these arrays up to g_AlphaSize
depending on a DNA or protein alignment. The fixed allocation of length
20 for each of these arrays is more than needed for DNA alignments and
can add up to gigabytes. A DNA only version of MUSCLE has been tested by
setting the length of these arrays to 4. It was able to align a large
data set on a desktop computer, whereas the stock MUSCLE crashed a 128
GB node on the university supercomputer.

Many of the other changes are in support of the above enhancements. For
example, only '-' is used internally as the gap character to improve the
performance of pairwise distance calculations from an MSA. When SSE code
is used, the increase is substantial. Gap characters other than '-' are
converted when reading the sequence files. Several source files have
been lightly touched to improve the profiling of different stages of
MUSCLE.

File: aligngivenpath.cpp
  Use '-' internally as the only gap character
  Initialize PPStart dynamically for SetXXX() functions
  Use g_AlphaSize instead of a constant

File: aligngivenpathsw.cpp
  Use '-' internally as the only gap character

File: aligntwoprofs.cpp
  Remove asserts and dead code

File: alpha.cpp
  Use '-' internally as the only gap character

File: alpha.h
  Use '-' internally as the only gap character

File: bittraceback.cpp
  * Add option to pack the traceback array in nibbles

File: cons.cpp
  * SSE acceleration of GetPctIdentityPair()

File: clust.cpp
  Improve timing and progress display

File: distcalc.cpp
  Remove unnecessary (float) type casts and conversions

File: distfunc.cpp
  * Fix memory leak

File: distpwkimura.cpp
  Progress fixes

File: domuscle.cpp
  Improve timing and progress display
  Calculate Other time in relation to Align time
  Time stages

File: estring.cpp
  * Use a typedef for the estring type

File: estring.h
  * Use a typedef for the estring type

File: fasta2.cpp
  Use '-' internally as the only gap character

File: fastclust.cpp
  Improve timing and progress display
  Time stages

File: fastdistjones.cpp
  Improve timing and progress display

File: fastdistmafft.cpp
  Improve timing and progress display

File: fastdistnuc.cpp
  Improve timing and progress display
  * Improve performance and simplify
  * SSE acceleration of CompareTuples()

File: glbalign.cpp
  Improve timing and progress display

File: globalslinux.cpp
  Use defined(__unix__) and fix some corner cases

File: globalsosx.cpp
  Remove dead code

File: globalsother.cpp
  Use defined(__unix__)

File: Makefile
  Specify Makefile in snapshot
  * Add make depend
  * Complete Makefile

File: makerootmsa.cpp
  * Use a typedef for the estring type

File: msa.cpp
  Delete arrays m_szSeqs and m_szNames

File: msa2.cpp
  * Divide ambiguous nucleotide weight by 4 instead of 20.
  * Add ambiguous residue weight to total weight.

File: msf.cpp
  Use '-' internally as the only gap character

File: muscle.h
  * Add option to pack the traceback array in nibbles
  * Change type of m_uSortOrder to unsigned char in struct ProfPos

File: nwdasmall.cpp
  Change the traceback array type to unsigned

File: nwsmall.cpp
  * Add option to pack the traceback array in nibbles

File: objscore.cpp
  Improve timing and progress display
  * Objective function score g_ticksObjScore counted twice

File: profile.h
  * Change type of m_uSortOrder to unsigned char in struct ProfPos
  * Use a typedef for the estring type

File: profilefrommsa.cpp
  * Change type of m_uSortOrder to unsigned char in struct ProfPos

File: progalign.cpp
  Improve timing and progress display
  Remove unnecessary argument from ProfileFromMSA()

File: progress.cpp
  Progress fixes
  Improve timing and progress display

File: progressivealign.cpp
  Improve timing and progress display

File: pwpath.cpp
  Remove dead code

File: realigndiffs.cpp
  Improve timing and progress display

File: realigndiffse.cpp
  Improve timing and progress display

File: refinehoriz.cpp
  Improve timing and progress display
  * Remove unused test code that causes a crash

File: refinesubfams.cpp
  Improve timing and progress display

File: refinetree.cpp
  Improve timing and progress display

File: refinetreee.cpp
  Improve timing and progress display

File: scoredist.cpp
  Progress fixes

File: scorehistory.cpp
  Improve timing and progress display

File: seq.cpp
  Use '-' internally as the only gap character

File: seqvect.cpp
  Use '-' internally as the only gap character

File: subfam.cpp
  Improve timing and progress display

File: timing.h
  Add timing options RDTSC, TIMEOFDAY, GETTIME

File: treefrommsa.cpp
  Improve timing and progress display

File: upgma2.cpp
  Improve timing and progress display