src.reverse()
X, Y = [0], [0]
x, y = 0, 0
for i in src:
if i[1] == 0:
y += 1
else:
x += 1
X.append(x)
Y.append(y)
return X, Y
################################################################################
posReadsPaths = getReadsInFolder(readsPosFilePath, minSize=0)
negReadsPaths = getReadsInFolder(readsNegFilePath, minSize=0)
index = getAlignedIndex(alignedSquiggles)
ref = Fasta(refFilePath)
# load basecalled sequence and signal
mod = KmerModel.load_from_hdf5(kmerModelFilePath)
################################################################################
pomery = [[[] for j in kmerLen] for i in levels]
overlap = [[[] for j in kmerLen] for i in levels]
goodDash = [[] for i in levels]
badDash = [[] for i in levels]
readsFilePath = "../../../data/pos-basecalled"
maxReads = 1000
import os
import sys
import numpy as np
sys.path.append("../")
from signalHelper import getReadsInFolder, getSignalFromRead, getSeqfromRead
reads = getReadsInFolder(readsFilePath, minSize=0)
signalLengths, seqLengths = [], []
for read in reads:
if maxReads == 0:
break
try:
signal = getSignalFromRead(read)
seq, _ = getSeqfromRead(read)
except:
continue
maxReads -= 1
signalLengths.append(len(signal))
seqLengths.append(len(seq))
meanSignal = np.mean(signalLengths)
len(signal),
signalShift,
signalScale,
levels,
overflow=overflow)
return signalString
referenceIdx = mp.Aligner(refFile)
assert referenceIdx, "failed to load/build reference index"
ref = Fasta(refFile)
mod = KmerModel.load_from_hdf5(kmerModelFilePath)
posReadsFiles = glob.glob(readsPosFilePath + '/*.fast5', recursive=True)
negReadsFiles = getReadsInFolder(readsNegFilePath)
contig = ref[0]
refSignal = stringToSignal(str(contig), mod, repeatSignal=repeatSignal)
refString = getGlobalString(refSignal)
helperDict = {"a": "A", "b": "C", "c": "G", "d": "T"}
#refString = "".join(helperDict[i] for i in refString)
#with open("./nieco.fa", "w") as f:
# f.write(">" + "x" + "\n")
# f.write(refString.upper() + "\n")
refLevelIdx = mp.Aligner("./helper.fa")
assert refLevelIdx, "failed to load/build reference index"
print("Priprava hotovo!")
if len(tempL) != 0:
dp[i] += max(tempL)
if max(dp) <= limit:
continue
#print(f"{x[begg:endd]}")
count += 1
out.append(x[begg][0])
print(f"Count is {count}")
return out
################################################################################
posReadsPaths = getReadsInFolder(readsPosFilePath, minSize=1000000)
referenceIdx = mp.Aligner(refFilePath)
assert referenceIdx, "failed to load/build reference index"
ref = Fasta(refFilePath)
# load basecalled sequence and signal
mod = KmerModel.load_from_hdf5(kmerModelFilePath)
################################################################################
globalNorms = {}
storeContig = {}
infoString = None
with open(refIndex, "r") as outFile: