| Change |
|
Comments |
|
Command line |
|
The
command line has been redesigned. It is not backwards
compatible with v5. I understand the importance of backwards
compatibility and regret the inconvenience this will cause for some users. This
was necessary because the underlying code has changed so much that simulating
the old command line was simply not possible. In v6, the command line is easier
to understand. Options are consistently supported across most commands, and
error messages are issued for invalid combinations of options. |
|
Faster and smaller |
|
Some commands are significantly faster and/or smaller
than their equivalents in v5. Smaller means less memory. These
improvements are data-dependent, so your performance may vary. In
rare cases, performance may be slightly worse. |
|
Multithreading |
|
Most search and clustering commands, including chimera-finding, now
support multithreading. This can give significant speed improvements on multi-core
CPUs. For example, on my 12-core i7 desktop PC, UCHIME is about 10x faster in
reference database mode. |
|
Identity definition |
|
In version 6, USEARCH always uses the BLAST definition of identity. |
|
Increased number of clusters |
|
The changed identity definition tends to increase the
number of clusters at a given identity compared with v5, but this
does not indicate reduced sensitivity. See cluster quality and sequence
identity. |
|
Database files |
|
Most search commands now support both FASTA and the
UDB database file format. This format
combines features of the old .udb and .wdb formats. Note that UDB files built
with v6 are not compatible with v5, and vice versa. In v6, UDB files are compatible
between 32- and 64-bit binaries. The database file is specified with the ‑db option;
the format (FASTA or UDB) is detected automatically. |
|
Obsoleted commands |
|
Some commands are no longer supported, including uhire
and ustar. The -msaout option replaces ustar. |