.\" Copyright (c) 2022 Stijn van Dongen
.TH "clmprotocols" 5 "9 Oct 2022" "clmprotocols 22-282" "FILE FORMATS "
.po 2m
.de ZI
.\" Zoem Indent/Itemize macro I.
.br
'in +\\$1
.nr xa 0
.nr xa -\\$1
.nr xb \\$1
.nr xb -\\w'\\$2'
\h'|\\n(xau'\\$2\h'\\n(xbu'\\
..
.de ZJ
.br
.\" Zoem Indent/Itemize macro II.
'in +\\$1
'in +\\$2
.nr xa 0
.nr xa -\\$2
.nr xa -\\w'\\$3'
.nr xb \\$2
\h'|\\n(xau'\\$3\h'\\n(xbu'\\
..
.if n .ll -2m
.am SH
.ie n .in 4m
.el .in 8m
..

.ZJ 3m 1m "1\&."
NAME
.in -4m
.ZJ 3m 1m "2\&."
DESCRIPTION
.in -4m
.ZJ 3m 1m "3\&."
Network representation
.in -4m
.ZJ 3m 1m "4\&."
Loading large networks
.in -4m
.ZJ 3m 1m "5\&."
Converting between formats
.in -4m
.ZJ 3m 1m "6\&."
Using threading and job dispatching
.in -4m
.ZJ 3m 1m "7\&."
Clustering similarity graphs encoded in BLAST results
.in -4m
.ZJ 3m 1m "8\&."
Clustering expression data
.in -4m
.ZJ 3m 1m "9\&."
Reducing node degrees in the graph
.in -4m
.ZJ 3m 1m "10\&."
SEE ALSO
.in -4m
.ZJ 3m 1m "11\&."
AUTHOR
.in -4m
.SH NAME
clmprotocols \- Work flows and protocols for mcl and friends
.SH DESCRIPTION
A guide to doing analysis with mcl and its helper programs\&.
.SH Network representation

The clustering program \fBmcl\fP expects the name of file as its first argument\&.
If the \fB--abc\fP option is used, the file is assumed to adhere to a
simple format where a network is specified edge by edge, one line and one
edge at a time\&.
Each line describes an edge as two labels and a numerical value, all
separated by white space\&. The labels and the value respectively identify the
two nodes and the edge weight\&. The format is called ABC-format,
where \&'A\&' and \&'B\&' represent the two labels and \&'C\&' represents the
edge weight\&. The latter is optional; if omitted the edge weight is set to one\&.
If ABC-format is used, the output is returned as a listing of clusters,
each cluster given as a line of white-space separated labels\&.

MCL can also utilize a second representation, which is a stringent and
unambiguous format for both input and output\&.
This is called \fImatrix format\fP and it is required when using other
programs in the mcl suite, for example when comparing and analysing
clusterings using \fBclm(1)\fP or when extracting and transforming
networks using \fBmcx(1)\fP\&.
Native mode (matrix format) is entered simply by \fInot\fP specifying
\fB--abc\fP\&.

The recommended approach using \fBmcl\fP is to convert an external format to
ABC-format\&. The program \fBmcxload(1)\fP reads the latter and creates a
native network file and a dictionary file that maps network nodes to
labels\&. All applications in the MCL suite, including \fBmcl\fP itself, can read
this native network file format\&. Label output can be obtained using
\fBmcxdump(1)\fP\&. The workflow is thus:

.di ZV
.in 0
.nf \fC
   #  External format has been converted to file data\&.abc (abc format)

   mcxload -abc data\&.abc --stream-mirror -write-tab data\&.tab -o data\&.mci  

   mcl data\&.mci -I 1\&.4
   mcl data\&.mci -I 2
   mcl data\&.mci -I 4

   mcxdump -icl out\&.data\&.mci\&.I14 -tabr data\&.tab -o dump\&.data\&.mci\&.I14
   mcxdump -icl out\&.data\&.mci\&.I20 -tabr data\&.tab -o dump\&.data\&.mci\&.I20
   mcxdump -icl out\&.data\&.mci\&.I40 -tabr data\&.tab -o dump\&.data\&.mci\&.I40
.fi \fR
.in
.di
.ne \n(dnu
.nf \fC
.ZV
.fi \fR

In this example the cluster output is stored in native format and dumped to
labels using mcxdump\&. The stored output can now be used to learn more about
the clusterings\&. An example is the following, where \fBclm(1)\fP is applied
in mode\ \&\fBdist\fP to gauge the distance between different clusterings\&.

.di ZV
.in 0
.nf \fC
   clm dist --chain out\&.data\&.mci\&.I{14,20,40}
.fi \fR
.in
.di
.ne \n(dnu
.nf \fC
.ZV
.fi \fR

.SH Loading large networks

If you deal with very large networks (say with hundreds of millions
of edges), it is recommended to use binary format (cf \fBmcxio(5)\fP)\&.
This is simply achieved by adding \fC--write-binary\fP to the mcxload
command line\&. The resulting file is no longer human-readable but
will be faster to read by a factor between ten- or twenty-fold
compared to standard MCL-edge network format, and a factor around fifty-fold
compared to label format\&.
All MCL-edge programs are able to read binary format, and speed
of reading will be somewhere in the order of millions of edges
per second, compared to, for example, roughly 100K edges
per second for label format\&.

Memory usage for mcxload can be lowered
by replacing the option \fC--stream-mirror\fP with \fC-ri\ \&max\fP\&.
.SH Converting between formats

\fBConverting label format to tabular format\fP
.br
Label format, two or three (including weight) columns:

.di ZV
.in 0
.nf \fC
   mcxload -abc data\&.abc --stream-mirror -write-tab data\&.tab -o data\&.mci
   mcxdump -imx data\&.mci -tab data\&.tab --dump-table
.fi \fR
.in
.di
.ne \n(dnu
.nf \fC
.ZV
.fi \fR

Simple Interaction File (SIF) format:

.di ZV
.in 0
.nf \fC
   mcxload -sif data\&.sif --stream-mirror -write-tab data\&.tab -o data\&.mci
   mcxdump -imx data\&.mci -tab data\&.tab --dump-table
.fi \fR
.in
.di
.ne \n(dnu
.nf \fC
.ZV
.fi \fR

These two examples are very similar, and differ only in the way the input to \fBmcxload\fP is specified\&.
.SH Using threading and job dispatching

The programs \fBmcxarray\fP, \fBmcx\ \&clcf\fP, \fBmcx\ \&ctty\fP,
\fBmcx\ \&diameter\fP and \fBclm\ \&info2\fP can all make use of both
threading and job dispatching\&. The clustering program \fBmcl\fP
can only use threading\&.

Instructing these programs to use threads is easy\&. It just requires
supplying \fB-t\fP\ \&\fB<num>\fP, e\&.g\&. use \fB-t\fP\ \&\fB4\fP to generate four threads\&.
It is only sensible to use \fC<num>\fP threads on a machine that has at least \fC<num>\fP CPUs\&.
It is additionally recommended that a threaded program has exclusive access to those CPUs
and does not have to contend with other jobs\&.

For the afore-mentioned programs it is additionally possible to split the computational
load over multiple machines\&. If \fC<N>\fP machines are available then \fC<N>\fP jobs should
be started\&. Each job should have an identical parameter \fB-J\fP\ \&\fBN\fP (e\&.g\&. \fB-J\fP\ \&\fB10\fP),
and varying parameters \fB-j\fP\ \&\fB0\fP, \fB-j\fP\ \&\fB1\fP, \&.\&.\&. \fB-j\fP\ \&\fBN-1\fP (e\&.g\&. \fB-j\fP\ \&\fB9\fP)\&.
It is possible to use threads in each individual job, but the number of threads should be
identical across all jobs issued\&. Output should typically be directed using a convention
such as \fB-o\fP\ \&\fBout\&.0\fP, \fB-o\fP\ \&\fBout\&.1\fP, \&.\&.\&. \fB-o\fP\ \&\fBout\&.9\fP\&.

After all jobs have finished the outputs must be combined to form the final answer\&.
The manner in which this is done is dependent on the program used\&.
With the example output above this would be done as follows\&. It can be seen
that \fBclm\ \&info2\fP is not yet supported by \fBmcx\ \&collect\fP and requires somewhat
idiosyncratic processing\&.

.di ZV
.in 0
.nf \fC
   # mcx diameter:
      mcx collect --add-column -o out\&.diameter out\&.{0,1,2,3,4,5,6,7,8,9}

   # mcx ctty:
      mcx collect --add-column -o out\&.ctty out\&.{0,1,2,3,4,5,6,7,8,9}

   # mcx clcf:
      mcx collect --add-column -o out\&.clcf out\&.{0,1,2,3,4,5,6,7,8,9}

   # mcxarray:
      mcx collect --add-matrix -o out\&.ctty out\&.{0,1,2,3,4,5,6,7,8,9}

   # clm info2:
      clxdo add_table out\&.{0,1,2,3,4,5,6,7,8,9} > out\&.info2
.fi \fR
.in
.di
.ne \n(dnu
.nf \fC
.ZV
.fi \fR

.SH Clustering similarity graphs encoded in BLAST results

A specific instance of the workflow above is the clustering of proteins based on
their sequence similarities\&. In the most typical scenario the external
format is BLAST output, which needs to be transformed to ABC format\&.
In the examples below the input is in columnar blast format
obtained with the blast -m8 option\&.
It requires a version of \fBmcl\fP at least as recent as \fC09-061\fP\&.
First we create an ABC-formatted file using the external columnar BLAST
format, which is assumed to be in a file called \fCseq\&.cblast\fP\&.

.di ZV
.in 0
.nf \fC
   cut -f 1,2,11 seq\&.cblast > seq\&.abc
.fi \fR
.in
.di
.ne \n(dnu
.nf \fC
.ZV
.fi \fR

The columnar format in the file \fCseq\&.cblast\fP has, for a given BLAST hit,
the sequence labels in the first two columns and the asssociated E-value in
column\ \&11\&. It is parsed by the standard UNIX cut(1) utility\&. The format
must have been created with the BLAST -m8 option so that no comment lines
are present\&. Alternatively these can be filtered out using grep\&.
The newly created \fCseq\&.abc\fP file is loaded by \fBmcxload(1)\fP,
which writes both a network file \fCseq\&.mci\fP and a dictionary
file \fCseq\&.tab\fP\&.

.di ZV
.in 0
.nf \fC
   mcxload -abc seq\&.abc --stream-mirror --stream-neg-log10 -stream-tf \&'ceil(200)\&'
         -o seq\&.mci -write-tab seq\&.tab
.fi \fR
.in
.di
.ne \n(dnu
.nf \fC
.ZV
.fi \fR

The \fC--stream-mirror\fP option ensures that the resulting network will be
undirected, as recommended when using \fBmcl\fP\&. Omitting this option would
result in a directed network as BLAST E-values generally differ between two
sequences\&. The default course of action for \fBmcxload(1)\fP is to use the
best value found between a pair of labels\&. The next option,
\fC--abc-neg-log10\fP transforms the numerical values in the input (the BLAST
E-values) by taking the logarithm in base\ \&10 and subsequently negating the
sign\&. Finally, the transformed values are capped so that any E-value below
1e-200 is set to a maximum allowed edge weight of\ \&200\&.

To obtain clusterings from \fCseq\&.mci\fP and \fCseq\&.tab\fP one has two
choices\&. The first is to generate an abstract clustering representation
and from that obtain the label output, as follows\&.
Below the \fB-o\fP option is not used, so mcl will create meaningful and
unique output names by itself\&. The default way of doing this is to preprend
the prefix \fCout\&.\fP and to append a suffix encoding the inflation value
used, with inflation encoded using two digits of precision and the decimal
separator removed\&.

.di ZV
.in 0
.nf \fC
   mcl seq\&.mci -I 1\&.4
   mcl seq\&.mci -I 2
   mcl seq\&.mci -I 4
   mcl seq\&.mci -I 6

   mcxdump -icl out\&.seq\&.mci\&.I14 -tabr seq\&.tab -o dump\&.seq\&.mci\&.I14
   mcxdump -icl out\&.seq\&.mci\&.I20 -tabr seq\&.tab -o dump\&.seq\&.mci\&.I20
   mcxdump -icl out\&.seq\&.mci\&.I40 -tabr seq\&.tab -o dump\&.seq\&.mci\&.I40
   mcxdump -icl out\&.seq\&.mci\&.I60 -tabr seq\&.tab -o dump\&.seq\&.mci\&.I60
.fi \fR
.in
.di
.ne \n(dnu
.nf \fC
.ZV
.fi \fR

Now the file \fCout\&.seq\&.tab\&.I14\fP and its associates can be used for example
to compute the distances between the encoded clusterings with
\fBclm dist\fP, to compute a set of strictly reconciled nested clusterings
with \fBclm order\fP, or to compute an efficiency criterion with
\fBclm info\fP\&.

Alternatively, label output can be obtained directly from \fBmcl\fP
as follows\&.

.di ZV
.in 0
.nf \fC
   mcl seq\&.mci -I 1\&.4  -use-tab seq\&.tab
   mcl seq\&.mci -I 2  -use-tab seq\&.tab
   mcl seq\&.mci -I 4  -use-tab seq\&.tab
   mcl seq\&.mci -I 6  -use-tab seq\&.tab
.fi \fR
.in
.di
.ne \n(dnu
.nf \fC
.ZV
.fi \fR

.SH Clustering expression data

The clustering of expression data constitutes another workflow\&. In this case the
external format usually is a tabular file format containing labels for genes
or probes and numerical values measuring the expression values or fold
changes across a series of conditions or experiments\&. Such tabular files can
be processed by \fBmcxarray(1)\fP, which comes installed with \fBmcl\fP\&. The
program computes correlations (either Pearson or Spearmann) between genes,
and creates an edge between genes if their correlation exceeds the specified
cutoff\&. From this \fBmcxarray(1)\fP creates both a network file and a
dictionary file\&. In the example below, the file \fCexpr\&.data\fP is
in tabular format with one row of column headers (e\&.g\&. tags for
experiments) and one column of row identifiers (e\&.g\&. probe or gene identifiers)\&.

.di ZV
.in 0
.nf \fC
   mcxarray -data expr\&.data -skipr 1 -skipc 1 -o expr\&.mci -write-tab expr\&.tab --pearson -co 0\&.7 -tf \&'abs(),add(-0\&.7)\&'
   
.fi \fR
.in
.di
.ne \n(dnu
.nf \fC
.ZV
.fi \fR

This uses the Pearson correlation, ignoring values below 0\&.7\&.
The remaining values in the interval \fC[0\&.7-1]\fP are remapped to the interval
\fC[0-0\&.3]\fP\&. This is recommended so that the edge weights will have
increased contrast between them, as \fBmcl\fP is affected by relative differences
(ratios) between edge weights rather than absolute differences\&. To illustrate
this, values\ \&0\&.75 and\ \&0\&.95 are mapped to\ \&0\&.05 and\ \&0\&.25, with respective
ratios\ \&0\&.79 and\ \&0\&.25\&.
The network file \fCexpr\&.mci\fP and the dictionary file \fCexpr\&.tab\fP can
now be used as before\&.

It is possible to investigate the effect of the correlation cutoff as follows\&.
First a network is generated at a very low threshold, and this network
is analysed using \fBmcxquery\fP\&.

.di ZV
.in 0
.nf \fC
   mcxarray -data expr\&.data -skipr 1 -skipc 1 -o expr20\&.mcx --write-binary --pearson -co 0\&.2 -tf \&'abs()\&'
   mcx query -imx expr20\&.mcx --vary-correlation
   
.fi \fR
.in
.di
.ne \n(dnu
.nf \fC
.ZV
.fi \fR

The output is in a tabular format describing the properties of the network
at increasing correlation thresholds\&. Examples are the size of the biggest
component, the number of orphan nodes (not connected to any other node), and
the mean and median node degrees\&.
A good way to choose the cutoff is to balance the number of singletons
and the median node degree\&. Both should preferably not be too high\&.
For example the number of orphan nodes should be
less than ten percent of the total number of nodes,
and the median node degree should be at most one hundred neighbours\&.
.SH Reducing node degrees in the graph
A good way to lower node degrees in a network is to require that
an edge is among the best \fIk\fP edges (those of highest weight) for
\fIboth\fP nodes incident to the edge, for some value of \fIk\fP\&. This is
achieved by using \fCknn(k)\fP in the argument to the \fB-tf\fP option to
mcl or \fBmcx\ \&alter\fP\&.
To give an example, a graph was formed on translations in Ensembl release 57 on 2\&.6M nodes\&.
The similarities were obtained from BLAST scores,
leading to a graph with a total edge count of 300M, with
best-connected nodes of degree respectively
11148, 9083, 9070, 9019 and 8988, and with mean node degree 233\&.
These degrees are unreasonable\&.
The graph was subjected to \fBmcx\ \&query\fP to investigate the effect of
varying k-NN parameters\&. A good heuristic is to choose a value
that does not significantly change the number of singletons in the input graph\&.
In the example it meant that \fB-tf\fP\ \&\fB\&'knn(160)\&'\fP was feasible, leading
to a mean node degree of 98\&.

A second approach to reduce node degrees is to employ the \fB-ceil-nb\fP option\&.
This ranks nodes by node degree, highest first\&. Nodes are considered
in order of rank, and edges of low weight are removed from the graph until
a node satisfies the node degree threshold specified by \fB-ceil-nb\fP\&.

.SH SEE ALSO
\fBmcxio(5)\fP\&.
.SH AUTHOR
Stijn van Dongen\&.