mcxdump.azm

\import{mcx.zmm}

\begin{pud::man}{

   {name}{mcxdump}
   {html_title}{The mcxdump manual}
   {author}{Stijn van Dongen}
   {section}{1}
   {synstyle}{long}
   {defstyle}{long}

   \man_share
}

\${html}{\"pud::man::maketoc"}

\sec{name}{NAME}
\NAME{mcxdump}{dump matrices, optionally map indices to labels}


\sec{synopsis}{SYNOPSIS}
\par{
   \mcxdump
   \synoptopt{-imx}{<fname>}{matrix file}
   \synoptopt{-icl}{<fname>}{cluster file to be dumped line-wise}
   \synoptopt{-tf}{<spec>}{apply unary transformations to input matrix}
   \synoptopt{-imx-cat}{<fname>}{concatenation matrix file}
   \synoptopt{-imx-tree}{<fname>}{concatenation cone file}
   \synoptopt{--skeleton}{read empty matrix, honour domains}
   \synoptopt{-o}{<fname>}{output file name ('-' for stdout)}
   \synoptopt{-digits}{<num>}{output precision}
   \synoptopt{-tab}{<fname>}{row/column tab (label) file}
   \synoptopt{-tabc}{<fname>}{column tab file}
   \synoptopt{-tabr}{<fname>}{row tab file}
   \synoptopt{--lazy-tab}{allow tab/domain mismatch}
   \synoptopt{--transpose}{work with the transpose}
   \synoptopt{--no-values}{omit values}
   \synoptopt{--omit-empty}{omit empty columns}
   \synoptopt{--no-loops}{omit loops}
   \synoptopt{--force-loops}{force loops}
   \synoptopt{--dump-pairs}{emit pairs per line}
   \synoptopt{--dump-table}{dump table format}
   \synoptopt{-dump-sif}{<tag>}{dump sif format}
   \synoptopt{-dump-sifx}{<tag>}{dump extended sif format with weights}
   \synoptopt{--dump-lines}{emit rows per line}
   \synoptopt{--dump-rlines}{omit leading identifier}
   \synoptopt{--dump-vlines}{add leading identifier values}
   \synoptopt{--dump-lead-off}{omit leading identifier}
   \synoptopt{--dump-lower}{dump lower part excluding diagonal}
   \synoptopt{--dump-loweri}{dump lower part including diagonal}
   \synoptopt{--dump-upper}{dump upper part excluding diagonal}
   \synoptopt{--dump-upperi}{dump upper part including diagonal}
   \synoptopt{--write-tabc}{dump tab file on column domain}
   \synoptopt{--write-tabr}{dump tab file on row domain}
   \synoptopt{--dump-domc}{dump column domain}
   \synoptopt{--dump-domr}{dump row domain}
   \synoptopt{-table-nfields}{<num>}{output first <num> fields}
   \synoptopt{-table-nlines}{<num>}{output first <num> lines}
   \synoptopt{--newick}{output newick format}
   \synoptopt{-newick}{[NBI]+}{exclude Number|Branch-length|Indent}
   \synoptopt{--write-matrix}{(deconcatenate) write matrices}
   \synoptopt{-split-stem}{<str>}{(deconcatenate) matrices file name stem}
   \synoptopt{-cat-max}{<num>}{(deconcatenate) write first <num> matrices}
   \synoptopt{-sep-value}{<str>}{node/value separator}
   \synoptopt{-sep-field}{<str>}{field separator}
   \synoptopt{-sep-lead}{<str>}{lead separator}
   \synoptopt{-sep-cat}{<str>}{concatenation separator}
   \synoptopt{-prefixc}{<str>}{prefix column indices with <str>}
   \synoptopt{-sort}{size-{ascending,descending}}{vector sort mode}
   \stdsynopt
   }


\sec{description}{DESCRIPTION}
\par{
   \mcxdump reads a data file satisfying the mcl input format
   (refer to \mysib{mcxio}).  It outputs a line-based format. The
   \useopt{--dump-pairs} option yields a single matrix entry per line,
   identified by the respective column and row identifiers (either index or
   label) separated by the field separator.
   The \useopt{--dump-lines} and \useopt{--dump-rlines} result in the
   joining of all row entries on a single line, separated by the field
   separator. For both formats, the matrix value corresponding with
   a particular entry is by default output as well.
   }

\par{
   \mcxdump can also act on files that contain concatenated
   matrices. Refer to the group of options headed by
   \genoptref{-imx-cat}{fname}.
   }

\sec{options}{OPTIONS}

\begin{itemize}{\mcx_itemopts}


\item{\defopt{-imx}{<fname>}{matrix file}}
\car{
   Input matrix.}

\item{\defopt{-icl}{<fname>}{cluster file}}
\car{
   This specifies the input matrix, and sets up a cluster-wise line-based label dump.
   This option is fully equivalent to the combination of
   \genopt{--dump-rlines} and \genopt{--no-values}.
   }

\item{\defopt{-tf}{<spec>}{apply unary transformations to input matrix}}
\car{
   Applies the specified transformation to the matrix before it is output.
   Refer to \mysib{mcxio} for a description of the transformation syntax.
   }

\item{\defopt{--transpose}{work with the transpose}}
\car{
   Work with the tranpsose of the input matrix.
   }

\item{\defopt{--skeleton}{read empty matrix, honour domains}}
\car{
   No entries are read, only domains.
   }

\item{\defopt{-o}{<fname>}{output file name}}
\car{
   Output stream. Use \v{-} for STDOUT.}

\item{\defopt{-digits}{<num>}{output precision}}
\car{
   Specify the precision to use in native interchange format.}

\item{\defopt{-tab}{<fname>}{row/column tab (label) file}}
\car{
   Substitute column indices and row indices by labels from the tab file.
   Since the same tab file is used for both, this implies that the matrix
   domains are identical.}

\item{\defopt{-tabc}{<fname>}{column tab file}}
\car{
   Substitute column indices by labels from the tab file.}

\item{\defopt{-tabr}{<fname>}{row tab file}}
\car{
   Substitute row indices by labels from the tab file.}

\item{\defopt{--lazy-tab}{allow tab/domain mismatch}}
\car{
   If used, the tab file domain(s) do not necessarily need to match
   the corresponding domain in the input matrix. Entries missing in
   the tab files will be replaced by a question mark.}

\item{\defopt{--no-values}{omit values}}
\car{
   Do not emit values.}

\item{\defopt{--omit-empty}{omit empty columns}}
\car{
   Do not output line data (with \genopt{--dump-table} or
   \genopt{--dump-lines} or related options) for those columns
   that are empty.
   }

\item{\defopt{--no-loops}{omit loops}}
\car{
   Do not output entries for which the row index equals the column index,
   if present.
   Applies only to matrices for which column and row domains are equal.}

\item{\defopt{--force-loops}{force loops}}
\car{
   For each column, force output of a row entry that matches the
   column index.
   Applies only to matrices for which column and row domains are equal.}

\items{
   {\defopt{--dump-pairs}{emit pairs per line}}
   {\defopt{-dump-sif}{<tag>}{dump sif format}}
   {\defopt{-dump-sifx}{<tag>}{dump extended sif format with weights}}
   {\defopt{--dump-lines}{emit rows per line}}
   {\defopt{--dump-rlines}{omit leading column node}}
   {\defopt{--dump-vlines}{add leading column values}}
   {\defopt{--dump-lead-off}{do not dump leading identifiers}}
   {\defopt{--dump-lower}{dump lower part excluding diagonal}}
   {\defopt{--dump-loweri}{dump lower part including diagonal}}
   {\defopt{--dump-upper}{dump upper part excluding diagonal}}
   {\defopt{--dump-upperi}{dump upper part including diagonal}}
}
\car{
   \genopt{--dump-pairs} is the default mode of output. Each matrix entry
   is output as a single pair of column-identifier and row-identifier per line,
   optionally followed by the value of the corresponding matrix entry.
   All fields are separated by the field separator.
   }

\par{
   Use \genopt{-dump-sif}{<tag>} to dump SIF format.
   The argument \genarg{<tag>} will be used as the edge type (the second
   column in SIF format). The option \genopt{-dump-sifx}{<tag>}
   is similar except that an extended format is produced where
   the label is followed by the colon character and the edge weight.
   }

\par{
   With \genopt{--dump-lines}, each matrix column is output on a
   single line, with row identifiers separated by the field separator
   and values attached to the row identifier by the node/value separator.
   In this format, the column identifier is output as the leading field.
   }

\par{
   \genopt{--dump-rlines} is as \genopt{--dump-lines},
   except that the column identifier is not output.
   Use \genopt{--dump-lead-off} to preclude the output of the leading
   identifiers (for line-based outputs).
   }

\par{
   \genopt{--dump-vlines} is as \genopt{--dump-lines}. The
   leading identifiers are followed by a value associated with
   the entire column. This can be used to dump the output
   given by \clmstable. The value provided is a measure
   for the stability of the cluster that follows.
   }

\par{
   The options pertaining to \it{lower} and \it{upper} dumps currently
   only work with \genopt{--dump-pairs}. They act to only output
   the specified part of the matrix.
   }

\items{
   {\defopt{--dump-table}{dump table format}}
   {\defopt{-table-nfields}{field limit}}
   {\defopt{-table-nlines}{line/row limit}}
}
\car{
   Output table format. In table format no indices are printed by default
   and all values
   are printed including zeroes.  The options \genopt{-table-nfields} and \genopt{-table-nlines}
   can be used to limit
   the number of fields and lines to be printed.  Note that fields correspond
   to MCL matrix rows and that lines correspond to MCL matrix columns, as MCL
   calls its primary indices column indices.
   Use \genopt{--dump-lead-off} to preclude the output of the leading
   identifiers (for line-based outputs).
   }

\items{
   {\defopt{--newick}{output newick format}}
   {\defopt{-newick}{[NBI]+}{newick, exclude Number|Branch-length|Indent}}
}
\car{
   Output a hierarchical clustering specified by \genopt{-imx-tree}
   in Newick tree format.
   }

\items{
   {\defopt{--write-tabc}{dump tab file on column domain}}
   {\defopt{--write-tabr}{dump tab file on row domain}}
   {\defopt{--dump-domc}{dump column domain}}
   {\defopt{--dump-domr}{dump row domain}}
}
\car{
   These options work in conjunction with the \genoptref{-ixm}{fname} option.
   Only the domains from the input matrix are read as if \genoptref{--skeleton}
   was specified.
   \genopt{--write-tabc} assumes the input tab file envelopes the matrix column
   domain, and it outputs a new tab file restricted to that domain.
   \genopt{--write-tabr} acts analogously for the row domain.
   \genopt{--dump-domc} and \genopt{--dump-domr} respectively dump the column
   or row domain as a regular dump, outputting labels in case a tab file is
   specified.
   }

\par{
   These options are implemented as ensembles of other options.
   For example, \genopt{--dump-domr} \genopt{-imx}{fname} corresponds with
   \genopt{--dump-lines} \genopt{--transpose} \genopt{--skeleton}.
   }

\items{
   {\defopt{-imx-cat}{<fname>}{concatenation matrix file}}
   {\defopt{-imx-tree}{<fname>}{concatenation cone file}}
   {\defopt{--write-matrix}{(deconcatenate) write matrices}}
   {\defopt{-split-stem}{<str>}{(deconcatenate) matrices file name stem}}
   {\defopt{-cat-max}{<num>}{(deconcatenate) write first <num> matrices}}
}
\car{
   \genopt{-imx-cat} is like \genopt{-imx} except that the input is assumed to
   contain multiple concatenated matrices.
   The matrices are dumped separated by the
   \it{cat separator} (cf. \genopt{-sep-cat}).
   Alternatively, the matrices can be written to different files
   using the \genopt{-split-stem} option.
   In this case it is possible to output each matrix in native format
   rather than as a dump by specifying \genopt{--write-matrix}.
   This makes mcxdump effectively act as a deconcatenator.
   In all cases (respectively dumping and writing matrices
   to either the same stream or multiple files) the number of
   matrices to be dumped can be limited with \genopt{-cat-max}.
   }

\par{
   \genopt{-imx-tree} is like \genopt{-imx-cat} except that the input
   is assumed to be in cone format (the format output by \mclcm).
   This format encodes a tree as a concatenation of matrices with
   nested domains. \mcxdump will project all levels of this tree
   so that all row domains are the same as the bottom row domain.
   This implies that a set of nested clusterings (on different node
   sets, as the set of clusters of a given level is the node set
   of the next level) is transformed
   into a set of flattened clusterings, all on the same node set.
   If you do not want this to happen, simply use \genopt{-imx-cat}.
   }

\item{\defopt{-sep-value}{<str>}{node/value separator}}
\car{
   Set the node/value separator for line based row ensemble output.}

\item{\defopt{-sep-field}{<str>}{field separator}}
\car{
   Set the field separator for different row indices in a given column.}

\item{\defopt{-sep-lead}{<str>}{lead separator}}
\car{
   Set the lead separator. In the \genopt{--dump-lines} format it
   separates the leading column index from the following ensembl of
   row indices. It can be useful to make this different from the
   field separator. One can for example grep for columns that have
   more than one entry in a matrix mapping nodes to clusters. This
   will find nodes in overlap.}

\item{\defopt{-sep-cat}{<str>}{concatenation separator}}
\car{
   Set the separator that is used between matrix dumps when a concatenation of
   matrices is dumped.}

\item{\defopt{-prefixc}{<str>}{prefix column indices with <str>}}
\car{
   This can be useful when external row names cannot be numbers and
   when a label dictionary is not available or not appropriate.}

\item{\defopt{-sort}{size-{ascending,descending}}{concatenation separator}}
\car{
   Reorder the matrix columns prior to dumping, based on the number of
   nonzero entries in each column.
   Do not use this in conjunction with a tab file for the column domain.}

\stddefopt

\end{itemize}

\sec{author}{AUTHOR}
\par{
   Stijn van Dongen.
   }

\sec{seealso}{SEE ALSO}

\par{
   \mysib{mcxload},
   \mysib{mcl},
   \mysib{mclfaq},
   and \mysib{mclfamily} for an overview of all the documentation
   and the utilities in the mcl family.
   }

\end{pud::man}