def main(argv=None):
if argv is None:
argv = sys.argv
parser = E.OptionParser(
version=
"%prog version: $Id: r_table2scatter.py 2782 2009-09-10 11:40:29Z andreas $"
)
parser.add_option(
"-c",
"--columns",
dest="columns",
type="string",
help=
"columns to take from table. Choices are 'all', 'all-but-first' or a ','-separated list of columns."
)
parser.add_option(
"--logscale",
dest="logscale",
type="string",
help="log-transform one or both axes [default=%Default].")
parser.add_option("-a",
"--hardcopy",
dest="hardcopy",
type="string",
help="write hardcopy to file [default=%default].",
metavar="FILE")
parser.add_option("-f",
"--file",
dest="input_filename",
type="string",
help="filename with table data [default=%default].",
metavar="FILE")
parser.add_option("-2",
"--file2",
dest="input_filename2",
type="string",
help="additional data file [default=%default].",
metavar="FILE")
parser.add_option(
"-s",
"--stats",
dest="statistics",
type="choice",
choices=("correlation", "spearman", "pearson", "count"),
help="statistical quantities to compute [default=%default]",
action="append")
parser.add_option("-p",
"--plot",
dest="plot",
type="choice",
choices=("scatter", "pairs", "panel", "bar",
"bar-stacked", "bar-besides", "1_vs_x",
"matched", "boxplot", "scatter+marginal",
"scatter-regression"),
help="plots to plot [default=%default]",
action="append")
parser.add_option(
"-t",
"--threshold",
dest="threshold",
type="float",
help="min threshold to use for counting method [default=%default].")
parser.add_option(
"-o",
"--colours",
dest="colours",
type="int",
help="column with colour information [default=%default].")
parser.add_option(
"-l",
"--plot-labels",
dest="labels",
type="string",
help="column labels for x and y in matched plots [default=%default].")
parser.add_option("-d",
"--add-diagonal",
dest="add_diagonal",
action="store_true",
help="add diagonal to plot [default=%default].")
parser.add_option("-e",
"--plot-legend",
dest="legend",
type="int",
help="column with legend [default=%default].")
parser.add_option("-r",
"--options",
dest="r_options",
type="string",
help="R plotting options [default=%default].")
parser.add_option("--format",
dest="format",
type="choice",
choices=("full", "sparse"),
help="output format [default=%default].")
parser.add_option("--title",
dest="title",
type="string",
help="""plot title [default=%default].""")
parser.add_option("",
"--xrange",
dest="xrange",
type="string",
help="x viewing range of plot [default=%default].")
parser.add_option("",
"--yrange",
dest="yrange",
type="string",
help="y viewing range of plot[default=%default].")
parser.add_option("--allow-empty-file",
dest="fail_on_empty",
action="store_false",
help="do not fail on empty input [default=%default].")
parser.add_option("--fail-on-empty",
dest="fail_on_empty",
action="store_true",
help="fail on empty input [default=%default].")
parser.set_defaults(hardcopy=None,
input_filename="",
input_filename2=None,
columns="all",
logscale=None,
statistics=[],
plot=[],
threshold=0.0,
labels="x,y",
colours=None,
diagonal=False,
legend=None,
title=None,
xrange=None,
yrange=None,
r_options="",
fail_on_empty=True,
format="full")
(options, args) = E.Start(parser)
if len(args) == 1 and not options.input_filename:
options.input_filename = args[0]
if options.columns not in ("all", "all-but-first"):
options.columns = [int(x) - 1 for x in options.columns.split(",")]
if options.colours:
options.colours -= 1
if options.legend:
options.legend -= 1
table = {}
headers = []
# read data matrix
if options.input_filename:
lines = IOTools.openFile(options.input_filename, "r").readlines()
else:
# note: this will not work for interactive viewing, but
# creating hardcopy plots works.
lines = sys.stdin.readlines()
lines = [x for x in lines if x[0] != "#"]
if len(lines) == 0:
if options.fail_on_empty:
raise IOError("no input")
E.warn("empty input")
E.Stop()
return
matrix, headers, colours, legend = readTable(lines,
"matrix",
take_columns=options.columns,
headers=True,
colours=options.colours,
row_names=options.legend)
if options.input_filename2:
# read another matrix (should be of the same format.
matrix2, headers2, colours2, legend2 = readTable(
lines,
"matrix2",
take_columns=options.columns,
headers=True,
colours=options.colours,
row_names=options.legend)
R.assign("headers", headers)
ndata = R("""length( matrix[,1] )""")[0]
if options.loglevel >= 1:
options.stdlog.write("# read matrix: %ix%i\n" % (len(headers), ndata))
if colours:
R.assign("colours", colours)
for method in options.statistics:
if method == "correlation":
cor = R.cor(matrix, use="pairwise.complete.obs")
writeMatrix(sys.stdout, cor, headers=headers, format="%5.2f")
elif method == "pearson":
options.stdout.write("\t".join(("var1", "var2", "coeff", "passed",
"pvalue", "n", "method",
"alternative")) + "\n")
for x in range(len(headers) - 1):
for y in range(x + 1, len(headers)):
try:
result = R("""cor.test( matrix[,%i], matrix[,%i] )""" %
(x + 1, y + 1))
except rpy.RPyException as msg:
E.warn(
"correlation not computed for columns %i(%s) and %i(%s): %s"
% (x, headers[x], y, headers[y], msg))
options.stdout.write(
"%s\t%s\t%s\t%s\t%s\t%i\t%s\t%s\n" %
(headers[x], headers[y], "na", "na", "na", 0, "na",
"na"))
else:
options.stdout.write(
"%s\t%s\t%6.4f\t%s\t%e\t%i\t%s\t%s\n" %
(headers[x], headers[y],
result.rx2('estimate').rx2('cor')[0],
Stats.getSignificance(
float(result.rx2('p.value')[0])),
result.rx2('p.value')[0],
result.rx2('parameter').rx2('df')[0],
result.rx2('method')[0],
result.rx2('alternative')[0]))
elif method == "spearman":
options.stdout.write("\t".join(("var1", "var2", "coeff", "passed",
"pvalue", "method",
"alternative")) + "\n")
for x in range(len(headers) - 1):
for y in range(x + 1, len(headers)):
result = R(
"""cor.test( matrix[,%i], matrix[,%i], method='spearman')"""
% (x + 1, y + 1))
options.stdout.write(
"%s\t%s\t%6.4f\t%s\t%e\t%i\t%s\t%s\n" %
(headers[x], headers[y], result['estimate']['rho'],
Stats.getSignificance(float(result['p.value'])),
result['p.value'], result['parameter']['df'],
result['method'], result['alternative']))
elif method == "count":
# number of shared elements > threshold
m, r, c = MatlabTools.ReadMatrix(open(options.input_filename, "r"),
take=options.columns,
headers=True)
mask = numpy.greater(m, options.threshold)
counts = numpy.dot(numpy.transpose(mask), mask)
writeMatrix(options.stdout, counts, headers=c, format="%i")
if options.plot:
# remove columns that are completely empty
if "pairs" in options.plot:
colsums = R('''colSums( is.na(matrix ))''')
take = [x for x in range(len(colsums)) if colsums[x] != ndata]
if take:
E.warn("removing empty columns %s before plotting" % str(take))
matrix = R.subset(matrix, select=[x + 1 for x in take])
R.assign("""matrix""", matrix)
headers = [headers[x] for x in take]
if legend:
legend = [headers[x] for x in take]
if options.r_options:
extra_options = ", %s" % options.r_options
else:
extra_options = ""
if options.legend is not None and len(legend):
extra_options += ", legend=c('%s')" % "','".join(legend)
if options.labels:
xlabel, ylabel = options.labels.split(",")
extra_options += ", xlab='%s', ylab='%s'" % (xlabel, ylabel)
else:
xlabel, ylabel = "", ""
if options.colours:
extra_options += ", col=colours"
if options.logscale:
extra_options += ", log='%s'" % options.logscale
if options.xrange:
extra_options += ", xlim=c(%f,%f)" % tuple(
map(float, options.xrange.split(",")))
if options.yrange:
extra_options += ", ylim=c(%f,%f)" % tuple(
map(float, options.yrange.split(",")))
if options.hardcopy:
if options.hardcopy.endswith(".eps"):
R.postscript(options.hardcopy)
elif options.hardcopy.endswith(".png"):
R.png(options.hardcopy, width=1024, height=768, type="cairo")
elif options.hardcopy.endswith(".jpg"):
R.jpg(options.hardcopy, width=1024, height=768, type="cairo")
for method in options.plot:
if ndata < 100:
point_size = "1"
pch = "o"
elif ndata < 1000:
point_size = "1"
pch = "o"
else:
point_size = "0.5"
pch = "."
if method == "scatter":
R("""plot( matrix[,1], matrix[,2], cex=%s, pch="o" %s)""" %
(point_size, extra_options))
if method == "scatter-regression":
R("""plot( matrix[,1], matrix[,2], cex=%s, pch="o" %s)""" %
(point_size, extra_options))
dat = R(
"""dat <- data.frame(x = matrix[,1], y = matrix[,2])""")
R("""new <- data.frame(x = seq( min(matrix[,1]), max(matrix[,1]), (max(matrix[,1]) - min(matrix[,1])) / 100))"""
)
mod = R("""mod <- lm( y ~ x, dat)""")
R("""predict(mod, new, se.fit = TRUE)""")
R("""pred.w.plim <- predict(mod, new, interval="prediction")"""
)
R("""pred.w.clim <- predict(mod, new, interval="confidence")"""
)
R("""matpoints(new$x,cbind(pred.w.clim, pred.w.plim[,-1]), lty=c(1,2,2,3,3), type="l")"""
)
R.mtext("y = %f * x + %f, r=%6.4f, n=%i" %
(mod["coefficients"]["x"],
mod["coefficients"]["(Intercept)"],
R("""cor( dat )[2]"""), ndata),
3,
cex=1.0)
elif method == "pairs":
if options.add_diagonal:
R("""panel.hist <- function( x,y,... ) { points(x,y,...); abline(0,1); }"""
)
else:
R("""panel.hist <- function( x,y,... ) { points(x,y,...); }"""
)
# There used to be a argument na_action="na.omit", but
# removed this as there appeared error messages saying
# "na.action is not a graphical parameter" and the
# plots showed occasionally the wrong scale.
# cex=point_size also caused trouble (error message:
# "X11 used font size 8 when 2 was requested" or
# similar)
if options.colours:
R.pairs(matrix,
pch=pch,
col=colours,
main=options.title,
panel="panel.hist",
labels=headers,
cex_labels=2.0)
else:
R.pairs(matrix,
pch=pch,
panel="panel.hist",
main=options.title,
labels=headers,
cex_labels=2.0)
elif method == "boxplot":
extra_options += ",main='%s'" % options.title
# set vertical orientation
if max([len(x) for x in headers]) > 40 / len(headers):
# remove xlabel:
extra_options = re.sub(", xlab='[^']+'", "", extra_options)
extra_options += ", names.arg=headers, las=2"
R("""op <- par(mar=c(11,4,4,2))"""
) # the 10 allows the names.arg below the barplot
R("""boxplot( matrix %s)""" % extra_options)
elif method == "bar" or method == "bar-stacked":
if not options.colours:
extra_options += ", col=rainbow(5)"
# set vertical orientation
if max([len(x) for x in headers]) > 40 / len(headers):
# remove xlabel:
extra_options = re.sub(", xlab='[^']+'", "", extra_options)
extra_options += ", names.arg=headers, las=2"
R("""op <- par(mar=c(11,4,4,2))"""
) # the 10 allows the names.arg below the barplot
R("""barplot(as.matrix(matrix), %s)""" % extra_options)
elif method == "bar-besides":
if not options.colours:
extra_options += ", col=rainbow(%i)" % ndata
# set vertical orientation
if max([len(x) for x in headers]) > 40 / len(headers):
# remove xlabel:
extra_options = re.sub(", xlab='[^']+'", "", extra_options)
extra_options += ", names.arg=headers, las=2"
R("""op <- par(mar=c(11,4,4,2))"""
) # the 10 allows the names.arg below the barplot
R("""barplot(as.matrix(matrix), beside=TRUE %s)""" %
extra_options)
elif method == "scatter+marginal":
if options.title:
# set the size of the outer margins - the title needs to be added at the end
# after plots have been created
R.par(oma=R.c(0, 0, 4, 0))
R("""matrix""")
R("""
x <- matrix[,1];
y <- matrix[,2];
xhist <- hist(x, breaks=20, plot=FALSE);
yhist <- hist(y, breaks=20, plot=FALSE);
top <- max(c(xhist$counts, yhist$counts));
nf <- layout(matrix(c(2,0,1,3),2,2,byrow=TRUE), c(3,1), c(1,3), respect=TRUE );
par(mar=c(3,3,1,1)) ;
plot(x, y, cex=%s, pch="o" %s) ;
par(mar=c(0,3,1,1)) ;
barplot(xhist$counts, axes=FALSE, ylim=c(0, top), space=0 ) ;
par(mar=c(3,0,1,1)) ;
title(main='%s');
barplot(yhist$counts, axes=FALSE, xlim=c(0, top), space=0, horiz=TRUE ) ;
title(main='%s');
""" % (point_size, extra_options, xlabel, ylabel))
if options.title:
R.mtext(options.title, 3, outer=True, line=1, cex=1.5)
elif method in ("panel", "1_vs_x", "matched"):
if method == "panel":
pairs = []
for x in range(len(headers) - 1):
for y in range(x + 1, len(headers)):
pairs.append((x, y))
elif method == "1_vs_x":
pairs = []
for x in range(1, len(headers)):
pairs.append((0, x))
# print matching columns
elif method == "matched":
pairs = []
for x in range(len(headers) - 1):
for y in range(x + 1, len(headers)):
if headers[x] == headers[y]:
pairs.append((x, y))
break
w = int(math.ceil(math.sqrt(len(pairs))))
h = int(math.ceil(float(len(pairs)) / w))
PosInf = 1e300000
NegInf = -1e300000
xlabel, ylabel = options.labels.split(",")
R("""layout(matrix(seq(1,%i), %i, %i, byrow = TRUE))""" %
(w * h, w, h))
for a, b in pairs:
new_matrix = [
x for x in zip(
list(matrix[a].values())[0],
list(matrix[b].values())[0])
if x[0] not in (float("nan"), PosInf, NegInf)
and x[1] not in (float("nan"), PosInf, NegInf)
]
try:
R("""plot(matrix[,%i], matrix[,%i], main='%s versus %s', cex=0.5, pch=".", xlab='%s', ylab='%s' )"""
% (a + 1, b + 1, headers[b], headers[a], xlabel,
ylabel))
except rpy.RException as msg:
print("could not plot %s versus %s: %s" %
(headers[b], headers[a], msg))
if options.hardcopy:
R['dev.off']()
E.info("matrix added as >matrix< in R.")
if not options.hardcopy:
if options.input_filename:
interpreter = code.InteractiveConsole(globals())
interpreter.interact()
else:
E.info(
"can not start new interactive session as input has come from stdin."
)
E.Stop()
def main(argv=None):
"""script main.
parses command line options in sys.argv, unless *argv* is given.
"""
if argv is None:
argv = sys.argv
parser = E.OptionParser(
version=
"%prog version: $Id: matrix2matrix.py 2782 2009-09-10 11:40:29Z andreas $"
)
parser.add_option("-m",
"--method",
dest="methods",
type="choice",
action="append",
choices=(
"normalize-by-min-diagonal",
"normalize-by-column",
"log",
"ln",
"negzero2value",
"set-diagonal",
"subtract-matrix",
"mix-matrix",
"normalize-by-matrix",
"normalize-by-column-max",
"normalize-by-row-max",
"normalize-by-column-min",
"normalize-by-row-min",
"normalize-by-column-median",
"normalize-by-row-median",
"normalize-by-column-mean",
"normalize-by-row-mean",
"normalize-by-column-total",
"normalize-by-row-total",
"correspondence-analysis",
"normalize-by-value",
"add-value",
"sort-rows",
"sort-columns",
"transpose",
"upper-bound",
"lower-bound",
"subtract-first-col",
"multiply-by-value",
"divide-by-value",
"mask-rows",
"mask-columns",
"mask-rows-and-columns",
"symmetrize-mean",
"symmetrize-max",
"symmetrize-min",
),
help="""method to use [default=%default]""")
parser.add_option("-s",
"--scale",
dest="scale",
type="float",
help="factor to scale matrix by [default=%default].")
parser.add_option("-f",
"--format",
dest="format",
type="string",
help="output number format [default=%default].")
parser.add_option("--filename-rows",
dest="filename_rows",
type="string",
help="filename with rows to mask [default=%default].")
parser.add_option("--filename-columns",
dest="filename_columns",
type="string",
help="filename with columns to mask [default=%default].")
parser.add_option("-p",
"--parameters",
dest="parameters",
type="string",
help="Parameters for various functions.")
parser.add_option("-t",
"--headers",
dest="headers",
action="store_true",
help="matrix has row/column headers.")
parser.add_option("--no-headers",
dest="headers",
action="store_false",
help="matrix has no row/column headers.")
parser.add_option("-a",
"--value",
dest="value",
type="float",
help="value to use for various algorithms.")
parser.add_option("-i",
"--input-format",
dest="input_format",
type="choice",
choices=("full", "sparse", "phylip"),
help="""input format for matrix.""")
parser.add_option("-o",
"--output-format",
dest="output_format",
type="choice",
choices=("full", "sparse", "phylip"),
help="""output format for matrix.""")
parser.add_option(
"--missing",
dest="missing",
type="float",
help=
"value to use for missing values. If not set, missing values will cause the script to fail [default=%default]."
)
parser.set_defaults(
methods=[],
scale=1.0,
headers=True,
format="%6.4f",
output_format="full",
input_format="full",
value=0.0,
parameters="",
write_separators=True,
filename_rows=None,
filename_columns=None,
missing=None,
)
(options, args) = E.Start(parser)
options.parameters = options.parameters.split(",")
lines = filter(lambda x: x[0] != "#", sys.stdin.readlines())
if len(lines) == 0:
raise IOError("no input")
chunks = filter(lambda x: lines[x][0] == ">", range(len(lines)))
if not chunks:
options.write_separators = False
chunks = [-1]
chunks.append(len(lines))
if options.filename_rows:
row_names, n = IOTools.ReadList(open(options.filename_rows, "r"))
if options.filename_columns:
column_names, n = IOTools.ReadList(open(options.filename_columns, "r"))
for chunk in range(len(chunks) - 1):
try:
raw_matrix, row_headers, col_headers = MatlabTools.readMatrix(
StringIO.StringIO("".join(lines[chunks[chunk] +
1:chunks[chunk + 1]])),
format=options.input_format,
headers=options.headers,
missing=options.missing)
except ValueError, msg:
E.warn("matrix could not be read: %s" % msg)
continue
nrows, ncols = raw_matrix.shape
E.debug("read matrix: %i x %i, %i row titles, %i colum titles" %
(nrows, ncols, len(row_headers), len(col_headers)))
parameter = 0
for method in options.methods:
matrix = numpy.reshape(numpy.array(raw_matrix), raw_matrix.shape)
if method in ("normalize-by-matrix", "subtract-matrix",
"mix-matrix", "add-matrix"):
other_matrix, other_row_headers, other_col_headers = MatlabTools.ReadMatrix(
open(options.parameters[parameter], "r"),
headers=options.headers)
other_nrows, other_ncols = other_matrix.shape
if options.loglevel >= 2:
options.stdlog.write(
"# read second matrix from %s: %i x %i, %i row titles, %i colum titles.\n"
% (options.parameters[parameter], other_nrows,
other_ncols, len(other_row_headers),
len(other_col_headers)))
parameter += 1
elif method == "normalize-by-min-diagonal":
for x in range(nrows):
for y in range(ncols):
m = min(raw_matrix[x, x], raw_matrix[y, y])
if m > 0:
matrix[x, y] = raw_matrix[x, y] / m
elif method == "normalize-by-column":
if nrows != ncols:
raise "only supported for symmeric matrices."
for x in range(nrows):
for y in range(ncols):
if raw_matrix[y, y] > 0:
matrix[x, y] = raw_matrix[x, y] / raw_matrix[y, y]
elif method == "normalize-by-value":
matrix = raw_matrix / float(options.parameters[parameter])
parameter += 1
elif method == "normalize-by-row":
if nrows != ncols:
raise "only supported for symmeric matrices."
for x in range(nrows):
for y in range(ncols):
if raw_matrix[y, y] > 0:
matrix[x, y] = raw_matrix[x, y] / raw_matrix[x, x]
elif method == "subtract-first-col":
for x in range(nrows):
for y in range(ncols):
matrix[x, y] -= raw_matrix[x, 0]
elif method.startswith("normalize-by-column"):
if method.endswith("max"):
f = max
elif method.endswith("min"):
f = min
elif method.endswith("median"):
f = scipy.median
elif method.endswith("mean"):
f = scipy.mean
elif method.endswith("total"):
f = sum
for y in range(ncols):
m = f(matrix[:, y])
if m != 0:
for x in range(nrows):
matrix[x, y] = matrix[x, y] / m
elif method.startswith("normalize-by-row"):
if method.endswith("max"):
f = max
elif method.endswith("min"):
f = min
elif method.endswith("median"):
f = scipy.median
elif method.endswith("mean"):
f = scipy.mean
elif method.endswith("total"):
f = sum
for x in range(nrows):
m = f(matrix[x, :])
if m != 0:
for y in range(ncols):
matrix[x, y] = raw_matrix[x, y] / m
elif method == "negzero2value":
# set zero/negative values to a value
for x in range(nrows):
for y in range(ncols):
if matrix[x, y] <= 0:
matrix[x, y] = options.value
elif method == "minmax":
# set zero/negative values to a value
for x in range(nrows):
for y in range(ncols):
matrix[x, y], matrix[y, x] = \
min(matrix[x, y], matrix[y, x]), \
max(matrix[x, y], matrix[y, x])
elif method == "log":
# apply log to all values.
for x in range(nrows):
for y in range(ncols):
if matrix[x, y] > 0:
matrix[x, y] = math.log10(matrix[x, y])
elif method == "ln":
for x in range(nrows):
for y in range(ncols):
if matrix[x, y] > 0:
matrix[x, y] = math.log(matrix[x, y])
elif method == "transpose":
matrix = numpy.transpose(matrix)
row_headers, col_headers = col_headers, row_headers
nrows, ncols = ncols, nrows
elif method == "mul":
matrix = numpy.dot(matrix, numpy.transpose(matrix))
col_headers = row_headers
elif method == "multiply-by-value":
matrix *= options.value
elif method == "divide-by-value":
matrix /= options.value
elif method == "add-value":
matrix += options.value
elif method == "angle":
# write angles between col vectors
v1 = numpy.sqrt(numpy.sum(numpy.power(matrix, 2), 0))
matrix = numpy.dot(numpy.transpose(matrix), matrix)
row_headers = col_headers
nrows = ncols
for x in range(nrows):
for y in range(ncols):
matrix[x, y] /= v1[x] * v1[y]
elif method == "euclid":
# convert to euclidean distance matrix
matrix = numpy.zeros((ncols, ncols), numpy.float)
for c1 in range(0, ncols - 1):
for c2 in range(c1 + 1, ncols):
for r in range(0, nrows):
d = raw_matrix[r][c1] - raw_matrix[r][c2]
matrix[c1, c2] += (d * d)
matrix[c2, c1] = matrix[c1, c2]
matrix = numpy.sqrt(matrix)
row_headers = col_headers
nrows = ncols
elif method.startswith("symmetrize"):
f = method.split("-")[1]
if f == "max":
f = max
elif f == "min":
f = min
elif f == "mean":
f = lambda x, y: float(x + y) / 2
if nrows != ncols:
raise ValueError(
"symmetrize only available for symmetric matrices")
if row_headers != col_headers:
raise ValueError(
"symmetrize not available for permuted matrices")
for x in range(nrows):
for y in range(ncols):
matrix[x, y] = matrix[y,
x] = f(matrix[x, y], matrix[y,
x])
elif method == "sub":
matrix = options.value - matrix
elif method in ("lower-bound", "upper-bound"):
boundary = float(options.parameters[parameter])
new_value = float(options.parameters[parameter + 1])
parameter += 2
if method == "upper-bound":
for x in range(nrows):
for y in range(ncols):
if matrix[x, y] > boundary:
matrix[x, y] = new_value
else:
for x in range(nrows):
for y in range(ncols):
if matrix[x, y] < boundary:
matrix[x, y] = new_value
elif method == "subtract-matrix":
matrix = matrix - other_matrix
elif method == "add-matrix":
matrix = matrix + other_matrix
elif method == "normalize-by-matrix":
# set 0s to 1 in the other matrix
for x in range(nrows):
for y in range(ncols):
if other_matrix[x, y] == 0:
other_matrix[x, y] = 1.0
matrix = matrix / other_matrix
elif method == "mix-matrix":
for x in range(len(other_row_headers) - 1):
for y in range(x + 1, len(other_col_headers)):
matrix[x, y] = other_matrix[x, y]
elif method == "set-diagonal":
value = float(options.parameters[parameter])
for x in range(min(nrows, ncols)):
matrix[x, x] = value
parameter += 1
elif method == "transpose":
matrix = numpy.transpose(raw_matrix)
row_headers, col_headers = col_headers, row_headers
elif method == "correspondence-analysis":
row_indices, col_indices = CorrespondenceAnalysis.GetIndices(
raw_matrix)
map_row_new2old = numpy.argsort(row_indices)
map_col_new2old = numpy.argsort(col_indices)
matrix, row_headers, col_headers = CorrespondenceAnalysis.GetPermutatedMatrix(
raw_matrix,
map_row_new2old,
map_col_new2old,
row_headers=row_headers,
col_headers=col_headers)
elif method == "mask-rows":
r = set(row_names)
for x in range(len(row_headers)):
if row_headers[x] in r:
matrix[x, :] = options.value
elif method == "mask-columns":
r = set(column_names)
for x in range(len(col_headers)):
if col_headers[x] in r:
matrix[:, x] = options.value
elif method == "mask-rows-and-columns":
r = set(row_names)
c = set(column_names)
for x in range(len(row_headers)):
for y in range(len(col_headers)):
if row_headers[x] in r and col_headers[y] in c:
matrix[x, y] = options.value
raw_matrix = numpy.reshape(numpy.array(matrix), matrix.shape)
else:
# for simple re-formatting jobs
matrix = raw_matrix
if options.write_separators:
options.stdout.write(lines[chunks[chunk]])
MatlabTools.writeMatrix(sys.stdout,
matrix,
value_format=options.format,
format=options.output_format,
row_headers=row_headers,
col_headers=col_headers)
for x in range(len(headers) - 1):
for y in range(x + 1, len(headers)):
result = R(
"""cor.test( matrix[,%i], matrix[,%i], method='spearman')"""
% (x + 1, y + 1))
options.stdout.write(
"%s\t%s\t%6.4f\t%s\t%e\t%i\t%s\t%s\n" %
(headers[x], headers[y], result['estimate']['rho'],
Stats.getSignificance(float(result['p.value'])),
result['p.value'], result['parameter']['df'],
result['method'], result['alternative']))
elif method == "count":
# number of shared elements > threshold
m, r, c = MatlabTools.ReadMatrix(open(options.input_filename, "r"),
take=options.columns,
headers=True)
mask = numpy.greater(m, options.threshold)
counts = numpy.dot(numpy.transpose(mask), mask)
writeMatrix(options.stdout, counts, headers=c, format="%i")
if options.plot:
# remove columns that are completely empty
if "pairs" in options.plot:
colsums = R('''colSums( is.na(matrix ))''')
take = [x for x in range(len(colsums)) if colsums[x] != ndata]
if take:
E.warn("removing empty columns %s before plotting" % str(take))
matrix = R.subset(matrix, select=[x + 1 for x in take])
R.assign("""matrix""", matrix)
