def compute_confidence_intervals(scaledContainmentsObserverved,
L,
k,
confidence,
s,
debug=False):
alpha = 1 - confidence
z_alpha = probit(1 - alpha / 2)
f1 = lambda Nm: 1 - 1.0 * Nm / L + z_alpha * sqrt(1.0 * Nm * (L - Nm) *
(1 - s) /
(s * L**3)) - Cks
f1_mpf = lambda Nm: mpf(f1(Nm))
f2 = lambda Nm: 1 - 1.0 * Nm / L - z_alpha * sqrt(1.0 * Nm * (L - Nm) *
(1 - s) /
(s * L**3)) - Cks
f2_mpf = lambda Nm: mpf(f2(Nm))
all_results = []
for (CksIx, Cks) in enumerate(scaledContainmentsObserverved):
sol1_mpf = brentq(f1_mpf, 0, L)
sol2_mpf = brentq(f2_mpf, 0, L)
sol1 = sol1_mpf
sol2 = sol2_mpf
Clow = 1 - 1.0 * sol1 / L
Chigh = 1 - 1.0 * sol2 / L
f3 = lambda pest: mpf((1 - pest)**k + z_alpha * sqrt(
thm5.var_n_mutated(L, k, pest)) / L - Clow)
f4 = lambda pest: mpf((1 - pest)**k - z_alpha * sqrt(
thm5.var_n_mutated(L, k, pest)) / L - Chigh)
#phigh = newton(f3, Clow)
#plow = newton(f4, Chigh)
phigh = brentq(f3, 0.0, 1.0)
plow = brentq(f4, 0.0, 1.0)
#print(phigh, f3(phigh))
#print(plow, f4(plow))
values = [L, k, confidence, Cks, Clow, Chigh, plow, phigh]
all_results.append(values)
return all_results
def var_c_scaled_one_step(L, k, p, s, confidence):
bias_factor = 1 - (1 - s)**L
var_multiplier = (1 - s)**2 / (L**6 * s**2 * bias_factor**2)
var_inner = lambda pest: L**2 * thm5.var_n_mutated(
L, k, pest) + var_n_mutated_squared(L, k, pest) - 2 * L * (
exp_n_mutated_cubed(L, k, pest) - exp_n_mutated(
L, k, pest) * exp_n_mutated_squared(L, k, pest))
var_direct = lambda pest: var_multiplier * var_inner(pest)
print(var_inner(p))
print(var_multiplier)
return var_direct(p)
def main():
global reportProgress, debug
# parse the command line
kmerSize = 28
kmerSequenceLength = 100
numSequences = None
noiseKind = None
pSubstitution = None
sequenceType = "linear"
confidence = 0.99
ciUseInverse = True
sortBy = "nMutated"
statsFilename = None
prngSeed = None
reportProgress = None
debug = []
statsOfInterest = [
"r1", "k", "L", "confidence", "trials", "q", "E[nMut].theory",
"StDev[nMut].theory", "inCI(r1est.nMut).obs", "Mean[nMut].obs",
"StDev[nMut].obs", "RMSE(StDev[nMut])", "RMSE(r1est.nMut)",
"E[nIsl].theory", "StDev[nIsl].theory", "inCI(r1est.nIsl).obs",
"Mean[nIsl].obs", "StDev[nIsl].obs", "RMSE(StDev[nIsl])",
"RMSE(r1est.nIsl)", "r1est.nIsl.impossible"
]
for arg in argv[1:]:
if ("=" in arg):
argVal = arg.split("=", 1)[1]
if (arg in ["--help", "-help", "--h", "-h"]):
usage()
elif (arg.startswith("--kmer=")) or (arg.startswith("K=")):
kmerSize = int(argVal)
elif (arg.startswith("--set=")) or (arg.startswith("N=")) or (
arg.startswith("L=")):
kmerSequenceLength = int_with_unit(argVal)
elif (arg.startswith("--sequences=")) or (arg.startswith("T=")):
numSequences = int_with_unit(argVal)
elif (arg.startswith("--poisson=")) or (
arg.startswith("--noise=")) or (arg.startswith("P=")):
noiseKind = "poisson"
pSubstitution = parse_probability(argVal)
elif (arg.startswith("--bernoulli=")) or (
arg.startswith("--error=")) or (arg.startswith("B=")) or (
arg.startswith("E=")):
noiseKind = "bernoulli"
pSubstitution = parse_probability(argVal)
elif (arg == "--linear"):
sequenceType = "linear"
elif (arg == "--circular"):
sequenceType = "circular"
elif (arg.startswith("--confidence=")) or (arg.startswith("C=")):
confidence = parse_probability(argVal)
elif (arg == "--noinverse"):
ciUseInverse = False
elif (arg == "--nosort"):
sortBy = None
elif (arg.startswith("--stats=")):
statsFilename = argVal
elif (arg.startswith("--seed=")):
prngSeed = argVal
elif (arg.startswith("--progress=")):
reportProgress = int_with_unit(argVal)
elif (arg == "--debug"):
debug += ["debug"]
elif (arg.startswith("--debug=")):
debug += argVal.split(",")
elif (arg.startswith("--")):
usage("unrecognized option: %s" % arg)
else:
usage("unrecognized option: %s" % arg)
if (pSubstitution == None):
usage("you have to tell me the mutation probability")
if (numSequences == None):
numSequences = 1
if (sequenceType == "circular"):
# all the estimator code assumes linear sequences
usage("circular sequences are not currently supported")
# set up randomness
if (prngSeed != None):
random_seed(prngSeed.encode("utf-8"))
if ("prng" in debug):
print("prng = %s" % prngSeed, file=stderr)
# set up model/generator
if (noiseKind == "poisson") and (sequenceType == "linear"):
mutationModel = PoissonModel \
(kmerSequenceLength+kmerSize-1,kmerSize,pSubstitution,
count_mutated_kmers_linear_naive if ("naive" in debug) else count_mutated_kmers_linear,
count_islands_linear)
elif (noiseKind == "bernoulli") and (sequenceType == "linear"):
mutationModel = BernoulliModel \
(kmerSequenceLength+kmerSize-1,kmerSize,pSubstitution,
count_mutated_kmers_linear_naive if ("naive" in debug) else count_mutated_kmers_linear,
count_islands_linear)
elif (noiseKind == "poisson") and (sequenceType == "circular"):
mutationModel = PoissonModel \
(kmerSequenceLength,kmerSize,pSubstitution,
count_mutated_kmers_circular_naive if ("naive" in debug) else count_mutated_kmers_circular,
count_islands_circular)
elif (noiseKind == "bernoulli") and (sequenceType == "circular"):
mutationModel = BernoulliModel \
(kmerSequenceLength,kmerSize,pSubstitution,
count_mutated_kmers_circular_naive if ("naive" in debug) else count_mutated_kmers_circular,
count_islands_circular)
else:
assert (False), "internal error"
# generate sequences and collect stats
nErrorsObserved = []
nMutatedObserved = []
nIslandObserved = []
r1EstNMutatedObserved = []
r1EstNIslandObserved = []
numImpossibleNislands = 0 # counts when nIsland can't be achieved with any r1
for seqNum in range(numSequences):
if (reportProgress != None):
if (1 + seqNum == 1) or ((1 + seqNum) % reportProgress == 0):
print("testing sequence %d" % (1 + seqNum), file=stderr)
# generate a (conceptual) sequence pair and collect stats
mutationModel.generate()
(nErrors, nMutated, nIsland) = mutationModel.count()
nErrorsObserved += [nErrors]
nMutatedObserved += [nMutated]
nIslandObserved += [nIsland]
r1EstNMutated = estimate_r1_from_n_mutated(kmerSequenceLength,
kmerSize, nMutated)
r1EstNMutatedObserved += [r1EstNMutated]
# note: when r1 is estimated from nIsland, there can be more than one
# solution; we take the solution that is closest to r1 estimated from
# nMutated
r1EstNIslandList = estimate_r1_from_n_island(kmerSequenceLength,
kmerSize, nIsland)
if (len(r1EstNIslandList) == 0):
r1EstNIsland = float("nan")
elif (len(r1EstNIslandList) == 1):
r1EstNIsland = r1EstNIslandList[0]
else:
r1EstNIsland = None
for r1Est in r1EstNIslandList:
diff = abs(r1Est - r1EstNMutated)
if (r1EstNIsland == None) or (diff < bestDiff):
r1EstNIsland = r1Est
bestDiff = diff
r1EstNIslandObserved += [r1EstNIsland]
if (impossible_n_island(kmerSequenceLength, kmerSize, nIsland)):
numImpossibleNislands += 1
# report per-trial results
if (sortBy == "nMutated"):
order = [(nMutatedObserved[ix], ix) for ix in range(numSequences)]
order.sort()
order = [ix for (_, ix) in order]
else: # if (sortBy == None):
order = list(range(numSequences))
header = [
"L", "K", "r", "trial", "nErr", "nMut", "nIsl", "r1est.nMut",
"r1.est.nIsl"
]
print("#%s" % "\t".join(header))
for ix in range(numSequences):
line = "\t".join(["%d","%d","%0.3f","%d","%d","%d","%d","%0.9f","%0.9f"]) \
% (kmerSequenceLength,
kmerSize,
pSubstitution,
1+order[ix],
nErrorsObserved[order[ix]],
nMutatedObserved[order[ix]],
nIslandObserved[order[ix]],
r1EstNMutatedObserved[order[ix]],
r1EstNIslandObserved[order[ix]])
print(line)
# compute stats
alpha = 1 - confidence
q = p_mutated(kmerSize, pSubstitution)
nMutatedMean = sample_mean(nMutatedObserved)
nMutatedStDev = sqrt(sample_variance(nMutatedObserved))
predNMutatedMean = exp_n_mutated(kmerSequenceLength, kmerSize,
pSubstitution)
predNMutatedStDev = sqrt(
var_n_mutated(kmerSequenceLength, kmerSize, pSubstitution))
rmseNMutatedStDev = abs(nMutatedStDev - predNMutatedStDev)
nIslandMean = sample_mean(nIslandObserved)
nIslandStDev = sqrt(sample_variance(nIslandObserved))
predNIslandMean = exp_n_island(kmerSequenceLength, kmerSize, pSubstitution)
predNIslandStDev = sqrt(
var_n_island(kmerSequenceLength, kmerSize, pSubstitution))
rmseNIslandStDev = abs(nIslandStDev - predNIslandStDev)
rmseR1EstNMutated = sqrt(
mean_squared_error(r1EstNMutatedObserved, pSubstitution))
rmseR1EstNIsland = sqrt(
mean_squared_error(r1EstNIslandObserved, pSubstitution))
(predR1EstNMutatedLow,predR1EstNMutatedHigh) \
= confidence_interval_r1_from_n_mutated(kmerSequenceLength,kmerSize,pSubstitution,alpha)
inConfR1EstNMutated \
= in_confidence_interval_q_from_n_mutated(kmerSequenceLength,kmerSize,pSubstitution,alpha,
nMutatedObserved,useInverse=ciUseInverse)
(predR1EstNIslandLow,predR1EstNIslandHigh) \
= confidence_interval_r1_from_n_island(kmerSequenceLength,kmerSize,pSubstitution,alpha)
inConfR1EstNIsland = in_confidence_interval_q_from_n_island(
kmerSequenceLength,
kmerSize,
pSubstitution,
alpha,
nIslandObserved,
nMutatedObserved,
useInverse=ciUseInverse)
# report stats
statToText = {}
statToText["r1"] = "%0.3f" % pSubstitution
statToText["k"] = "%d" % kmerSize
statToText["L"] = "%d" % kmerSequenceLength
statToText["confidence"] = "%0.3f" % confidence
statToText["trials"] = "%d" % numSequences
statToText["q"] = "%0.9f" % q
statToText["E[nMut].theory"] = "%0.9f" % predNMutatedMean
statToText["StDev[nMut].theory"] = "%0.9f" % predNMutatedStDev
statToText["CIlow(r1est.nMut).theory"] = "%0.9f" % predR1EstNMutatedLow
statToText["CIhigh(r1est.nMut).theory"] = "%0.9f" % predR1EstNMutatedHigh
statToText["inCI(r1est.nMut).obs"] = "%0.9f" % (
float(inConfR1EstNMutated) / numSequences)
statToText["Mean[nMut].obs"] = "%0.9f" % nMutatedMean
statToText["StDev[nMut].obs"] = "%0.9f" % nMutatedStDev
statToText["RMSE(StDev[nMut])"] = "%0.9f" % rmseNMutatedStDev
statToText["RMSE(r1est.nMut)"] = "%0.9f" % rmseR1EstNMutated
statToText["E[nIsl].theory"] = "%0.9f" % predNIslandMean
statToText["StDev[nIsl].theory"] = "%0.9f" % predNIslandStDev
statToText["CIlow(r1est.nIsl).theory"] = "%0.9f" % predR1EstNIslandLow
statToText["CIhigh(r1est.nIsl).theory"] = "%0.9f" % predR1EstNIslandHigh
statToText["inCI(r1est.nIsl).obs"] = "%0.9f" % (float(inConfR1EstNIsland) /
numSequences)
statToText["Mean[nIsl].obs"] = "%0.9f" % nIslandMean
statToText["StDev[nIsl].obs"] = "%0.9f" % nIslandStDev
statToText["RMSE(StDev[nIsl])"] = "%0.9f" % rmseNIslandStDev
statToText["RMSE(r1est.nIsl)"] = "%0.9f" % rmseR1EstNIsland
statToText["r1est.nIsl.impossible"] = "%0.9f" % (
float(numImpossibleNislands) / numSequences)
if (statsFilename != None):
if (statsFilename.endswith(".gz")) or (
statsFilename.endswith(".gzip")):
statsF = gzip_open(statsFilename, "wt")
else:
statsF = open(statsFilename, "wt")
print("#%s" % "\t".join(statsOfInterest), file=statsF)
statsLine = [statToText[stat] for stat in statsOfInterest]
print("\t".join(statsLine), file=statsF)
statsF.close()
else:
statW = max(len(stat) for stat in statsOfInterest)
for stat in statsOfInterest:
print("%*s = %s" % (stat, statW, statToText[stat]), file=stderr)
def main():
global reportProgress, debug
# parse the command line
kmerSize = 28
numSequences = None
noiseKind = None
pSubstitution = None
sequenceType = "linear"
confidence = 0.99
ciUseInverse = True
sortBy = "nMutated"
statsFilename = None
mutatedFilename = None
mutateOnly = False
prngSeed = None
reportProgress = None
debug = []
statsOfInterest = [
"name", "r1", "k", "L", "confidence", "trials", "q", "Mean[|A|].obs",
"Mean[|B|].obs", "Mean[|A^B|].obs", "Mean[|AuB|].obs",
"Mean[nMut.A,B].obs", "Mean[L.A,B].obs", "Mean[r1est.A,B].obs",
"inCI(r1est.A,B).obs"
]
for arg in argv[1:]:
if ("=" in arg):
argVal = arg.split("=", 1)[1]
if (arg in ["--help", "-help", "--h", "-h"]):
usage()
elif (arg.startswith("--kmer=")) or (arg.startswith("K=")):
kmerSize = int(argVal)
elif (arg.startswith("--sequences=")) or (arg.startswith("T=")):
numSequences = int_with_unit(argVal)
elif (arg.startswith("--poisson=")) or (
arg.startswith("--noise=")) or (arg.startswith("P=")):
noiseKind = "poisson"
pSubstitution = parse_probability(argVal)
elif (arg.startswith("--bernoulli=")) or (
arg.startswith("--error=")) or (arg.startswith("B=")) or (
arg.startswith("E=")):
noiseKind = "bernoulli"
pSubstitution = parse_probability(argVal)
elif (arg == "--linear"):
sequenceType = "linear"
elif (arg == "--circular"):
sequenceType = "circular"
elif (arg.startswith("--confidence=")) or (arg.startswith("C=")):
confidence = parse_probability(argVal)
elif (arg == "--noinverse"):
ciUseInverse = False
elif (arg == "--nosort"):
sortBy = None
elif (arg.startswith("--stats=")):
statsFilename = argVal
elif (arg.startswith("--mutated=")):
mutatedFilename = argVal
elif (arg == "--mutateonly"):
mutateOnly = True
elif (arg.startswith("--seed=")):
prngSeed = argVal
elif (arg.startswith("--progress=")):
reportProgress = int_with_unit(argVal)
elif (arg == "--debug"):
debug += ["debug"]
elif (arg.startswith("--debug=")):
debug += argVal.split(",")
elif (arg.startswith("--")):
usage("unrecognized option: %s" % arg)
else:
usage("unrecognized option: %s" % arg)
if (pSubstitution == None):
usage("you have to tell me the mutation probability")
if (numSequences == None):
numSequences = 1
if (noiseKind == "bernoulli"):
# the presence of non-ACGT nucleotides isn't considered
usage("the bernoulli noise model is not currently supported")
if (sequenceType == "circular"):
# all the estimator code assumes linear sequences
usage("circular sequences are not currently supported")
# set up randomness
if (prngSeed != None):
random_seed(prngSeed.encode("utf-8"))
# open a file to receive the mutated sequences
mutatedF = None
if (mutateOnly) and (mutatedFilename == None):
mutatedF = stdout
else:
if (mutatedFilename != None):
if (mutatedFilename.endswith(".gz")) or (
mutatedFilename.endswith(".gzip")):
mutatedF = gzip_open(mutatedFilename, "wt")
else:
mutatedF = open(mutatedFilename, "wt")
# fetch the *single* input sequence
numSequencesSeen = 0
for (seqName, seq) in fasta_sequences(stdin):
numSequencesSeen += 1
assert (numSequencesSeen <
2), "there was more than one sequence in the input"
seqLen = len(seq)
assert (numSequencesSeen == 1), "there were no sequences in the input"
ntSequenceLength = len(seq)
assert (
ntSequenceLength >= kmerSize
), "input sequence length (%d) is shorter than the kmer size (%d)" % (
ntSequenceLength, kmerSize)
distinctKmersA = kmer_set(seq, kmerSize)
numDistinctKmersA = len(distinctKmersA)
# set up model/generator
if (noiseKind == "poisson") and (sequenceType == "linear"):
kmerSequenceLength = ntSequenceLength - (kmerSize - 1)
mutationModel = PoissonModel \
(seq,kmerSize,pSubstitution,
count_mutated_kmers_linear)
elif (noiseKind == "bernoulli") and (sequenceType == "linear"):
kmerSequenceLength = ntSequenceLength - (kmerSize - 1)
mutationModel = BernoulliModel \
(seq,kmerSize,pSubstitution,
count_mutated_kmers_linear)
elif (noiseKind == "poisson") and (sequenceType == "circular"):
kmerSequenceLength = ntSequenceLength
mutationModel = PoissonModel \
(seq,kmerSize,pSubstitution,
count_mutated_kmers_circular)
elif (noiseKind == "bernoulli") and (sequenceType == "circular"):
kmerSequenceLength = ntSequenceLength
mutationModel = BernoulliModel \
(seq,kmerSize,pSubstitution,
count_mutated_kmers_circular)
else:
assert (False), "internal error"
# generate sequences and collect stats
alpha = 1 - confidence
nErrorsObserved = []
nMutatedObserved = []
r1EstNMutatedObserved = []
nDistinctAObserved = []
nDistinctBObserved = []
nDistinctIntersectionObserved = []
nDistinctUnionObserved = []
nMutatedABObserved = []
kmerSequenceLengthABObserved = []
r1EstABObserved = []
inConfR1EstABObserved = []
for seqNum in range(numSequences):
if (reportProgress != None):
if (1 + seqNum == 1) or ((1 + seqNum) % reportProgress == 0):
print("testing sequence %d" % (1 + seqNum), file=stderr)
# generate a mutated sequence and collect stats
mutatedSeq = mutationModel.generate()
if (mutatedF != None):
write_fasta(mutatedF, seqName + "_mutation_%d)" % (1 + seqNum),
mutatedSeq)
(nErrors, nMutated) = mutationModel.count()
nErrorsObserved += [nErrors]
nMutatedObserved += [nMutated]
r1EstNMutated = estimate_r1_from_n_mutated(kmerSequenceLength,
kmerSize, nMutated)
r1EstNMutatedObserved += [r1EstNMutated]
distinctKmersB = kmer_set(mutatedSeq, kmerSize)
numDistinctKmersB = len(distinctKmersB)
nDistinctKmersIntersection = len(
distinctKmersA.intersection(distinctKmersB))
nDistinctKmersUnion = len(distinctKmersA.union(distinctKmersB))
nDistinctAObserved += [numDistinctKmersA]
nDistinctBObserved += [numDistinctKmersB]
nDistinctIntersectionObserved += [nDistinctKmersIntersection]
nDistinctUnionObserved += [nDistinctKmersUnion]
kmerSequenceLengthAB = (numDistinctKmersA + numDistinctKmersB) / 2.0
nMutatedAB = kmerSequenceLengthAB - nDistinctKmersIntersection
r1EstAB = estimate_r1_from_n_mutated(kmerSequenceLengthAB, kmerSize,
nMutatedAB)
nMutatedABObserved += [nMutatedAB]
kmerSequenceLengthABObserved += [kmerSequenceLengthAB]
r1EstABObserved += [r1EstAB]
inConfR1EstAB = in_confidence_interval_q_from_n_mutated(
kmerSequenceLengthAB,
kmerSize,
pSubstitution,
alpha,
nMutatedAB,
useInverse=ciUseInverse)
inConfR1EstABObserved += [inConfR1EstAB]
# report per-trial results
if (sortBy == "nMutated"):
order = [(nDistinctIntersectionObserved[ix], ix)
for ix in range(numSequences)]
order.sort()
order.reverse()
order = [ix for (_, ix) in order]
else: # if (sortBy == None):
order = list(range(numSequences))
header = [
"L", "K", "r", "trial", "nErr", "nMut", "r1est.nMut", "|A|", "|B|",
"|A^B|", "|AuB|", "nMut.A,B", "L.A,B", "r1est.A,B", "inCI(r1est.A,B)"
]
print("#%s" % "\t".join(header))
for ix in range(numSequences):
line = "\t".join(["%d","%d","%0.3f","%d","%d","%d","%0.9f","%d","%d","%d","%d","%0.1f","%0.1f","%0.9f","%d"]) \
% (kmerSequenceLength, # L
kmerSize, # K
pSubstitution, # r
1+order[ix], # trial
nErrorsObserved[order[ix]], # nErr
nMutatedObserved[order[ix]], # nMut
r1EstNMutatedObserved[order[ix]], # r1est.nMut
nDistinctAObserved[order[ix]], # |A|
nDistinctBObserved[order[ix]], # |B|
nDistinctIntersectionObserved[order[ix]], # |A^B|
nDistinctUnionObserved[order[ix]], # |AuB|
nMutatedABObserved[order[ix]], # nMut.A,B
kmerSequenceLengthABObserved[order[ix]], # L.A,B
r1EstABObserved[order[ix]], # r1est.A,B
inConfR1EstABObserved[order[ix]]) # inCI(r1est.A,B)"]
print(line)
if (mutatedF != None) and (mutatedF != stdout):
mutatedF.close()
if (mutateOnly):
exit()
# compute stats
q = p_mutated(kmerSize, pSubstitution)
nMutatedMean = sample_mean(nMutatedObserved)
nMutatedStDev = sqrt(sample_variance(nMutatedObserved))
predNMutatedMean = exp_n_mutated(kmerSequenceLength, kmerSize,
pSubstitution)
predNMutatedStDev = sqrt(
var_n_mutated(kmerSequenceLength, kmerSize, pSubstitution))
rmseNMutatedStDev = abs(nMutatedStDev - predNMutatedStDev)
rmseR1EstNMutated = sqrt(
mean_squared_error(r1EstNMutatedObserved, pSubstitution))
(predR1EstNMutatedLow,predR1EstNMutatedHigh) \
= confidence_interval_r1_from_n_mutated(kmerSequenceLength,kmerSize,pSubstitution,alpha)
inConfR1EstNMutated = in_confidence_interval_q_from_n_mutated(
kmerSequenceLength,
kmerSize,
pSubstitution,
alpha,
nMutatedObserved,
useInverse=ciUseInverse)
nDistinctAMean = sample_mean(nDistinctAObserved)
nDistinctBMean = sample_mean(nDistinctBObserved)
nDistinctIntersectionMean \
= sample_mean(nDistinctIntersectionObserved)
nDistinctUnionMean = sample_mean(nDistinctUnionObserved)
nMutatedABMean = sample_mean(nMutatedABObserved)
kmerSequenceLengthABMean \
= sample_mean(kmerSequenceLengthABObserved)
r1EstABMean = sample_mean(r1EstABObserved)
# report stats
statToText = {}
statToText["name"] = seqName
statToText["r1"] = "%0.3f" % pSubstitution
statToText["k"] = "%d" % kmerSize
statToText["L"] = "%d" % kmerSequenceLength
statToText["confidence"] = "%0.3f" % confidence
statToText["trials"] = "%d" % numSequences
statToText["q"] = "%0.9f" % q
statToText["E[nMut].theory"] = "%0.9f" % predNMutatedMean
statToText["StDev[nMut].theory"] = "%0.9f" % predNMutatedStDev
statToText["CIlow(r1est.nMut).theory"] = "%0.9f" % predR1EstNMutatedLow
statToText["CIhigh(r1est.nMut).theory"] = "%0.9f" % predR1EstNMutatedHigh
statToText["inCI(r1est.nMut).obs"] = "%0.9f" % (
float(inConfR1EstNMutated) / numSequences)
statToText["Mean[nMut].obs"] = "%0.9f" % nMutatedMean
statToText["StDev[nMut].obs"] = "%0.9f" % nMutatedStDev
statToText["RMSE(StDev[nMut])"] = "%0.9f" % rmseNMutatedStDev
statToText["RMSE(r1est.nMut)"] = "%0.9f" % rmseR1EstNMutated
statToText["Mean[|A|].obs"] = "%d" % nDistinctAMean
statToText["Mean[|B|].obs"] = "%d" % nDistinctBMean
statToText["Mean[|A^B|].obs"] = "%d" % nDistinctIntersectionMean
statToText["Mean[|AuB|].obs"] = "%d" % nDistinctUnionMean
statToText["Mean[nMut.A,B].obs"] = "%d" % nMutatedABMean
statToText["Mean[L.A,B].obs"] = "%d" % kmerSequenceLengthABMean
statToText["Mean[r1est.A,B].obs"] = "%0.9f" % r1EstABMean
statToText["inCI(r1est.A,B).obs"] = "%0.9f" % (
float(sum(inConfR1EstABObserved)) / numSequences)
if (statsFilename != None):
if (statsFilename.endswith(".gz")) or (
statsFilename.endswith(".gzip")):
statsF = gzip_open(statsFilename, "wt")
else:
statsF = open(statsFilename, "wt")
print("#%s" % "\t".join(statsOfInterest), file=statsF)
statsLine = [statToText[stat] for stat in statsOfInterest]
print("\t".join(statsLine), file=statsF)
statsF.close()
else:
statW = max(len(stat) for stat in statsOfInterest)
for stat in statsOfInterest:
print("%*s = %s" % (stat, statW, statToText[stat]), file=stderr)
def fourth_moment_using_normal(L, k, p):
mu = exp_n_mutated(L, k, p)
var = var_n_mutated(L, k, p)
fourth_moment = mu**4 + 6 * mu**2 * var + 3 * var**2
return fourth_moment
def third_moment_nmut_using_normal(L, k, p):
mu = exp_n_mutated(L, k, p)
var = var_n_mutated(L, k, p)
third_moment = mu**3 + 3 * mu * var
return third_moment
def exp_n_mutated_squared(L, k, p):
return var_n_mutated(L, k, p) + exp_n_mutated(L, k, p)**2
def var_test(L, k, p, s, confidence):
print('test')
print(thm5.var_n_mutated(L, k, p) / L**2)