diff --git a/reacnetgenerator/_detect.py b/reacnetgenerator/_detect.py
index 88dc50935..db89ef741 100644
--- a/reacnetgenerator/_detect.py
+++ b/reacnetgenerator/_detect.py
@@ -140,8 +140,15 @@ def _readstepfunc(self, item):
         pass
 
     def _connectmolecule(self, bond, level):
+        # return list([b''.join((listtobytes(mol),
+        #                        listtobytes(bondlist))) for mol, bondlist in zip(*dps(bond, level))])
+        # int() because sometimes, the elements in bondlist are type numpy.int64 sometimes are int
+        # which cause the different bytes results from same bond-network.
         return list([b''.join((listtobytes(mol),
-                               listtobytes(bondlist))) for mol, bondlist in zip(*dps(bond, level))])
+                               listtobytes([(int(i[0]), int(i[1]))
+                                           for i in bondlist]).replace(b'\n', b'\t'),
+                               listtobytes([int(i[2]) for i in bondlist])
+                               )) for mol, bondlist in zip(*dps(bond, level))])
 
     def _writemoleculetempfile(self, d):
         with WriteBuffer(tempfile.NamedTemporaryFile('wb', delete=False)) as f:
@@ -328,7 +335,7 @@ def _readstepfunc(self, item):
                         s = line.split()
                         ss.append(list(map(float, s)))
         ss = np.array(ss)
-        if ss.shape[1]>2:
+        if ss.shape[1] > 2:
             xy = ss[0][2]
             xz = ss[1][2]
             yz = ss[2][2]
@@ -339,10 +346,10 @@ def _readstepfunc(self, item):
         ylo = ss[1][0] - min(0., yz)
         yhi = ss[1][1] - max(0., yz)
         zlo = ss[2][0]
-        zhi = ss[2][1] 
+        zhi = ss[2][1]
         boxsize = np.array([[xhi-xlo, 0., 0.],
-                   [xy, yhi-ylo, 0.],
-                   [xz, yz, zhi-zlo]])
+                            [xy, yhi-ylo, 0.],
+                            [xz, yz, zhi-zlo]])
         _, step_atoms = zip(*sorted(step_atoms, key=operator.itemgetter(0)))
         step_atoms = Atoms(step_atoms)
         bond, level = self._getbondfromcrd(step_atoms, boxsize)
@@ -355,7 +362,8 @@ class _Detectxyz(_DetectCrd):
     """xyz file. Two frames are connected. Cell information must be inputed by user."""
 
     def _readNfunc(self, f):
-        atomname_dict = dict(zip(self.atomname.tolist(), range(self.atomname.size)))
+        atomname_dict = dict(
+            zip(self.atomname.tolist(), range(self.atomname.size)))
         for index, line in enumerate(f):
             s = line.split()
             if index == 0:
@@ -380,7 +388,8 @@ def _readstepfunc(self, item):
         for index, line in enumerate(lines):
             s = line.split()
             if index > 1:
-                step_atoms.append((index-1, Atom(s[0] ,tuple((float(x) for x in s[1:4])))))
+                step_atoms.append(
+                    (index-1, Atom(s[0], tuple((float(x) for x in s[1:4])))))
         _, step_atoms = zip(*sorted(step_atoms, key=operator.itemgetter(0)))
         step_atoms = Atoms(step_atoms)
         bond, level = self._getbondfromcrd(step_atoms, boxsize)
diff --git a/reacnetgenerator/_mergeiso.py b/reacnetgenerator/_mergeiso.py
new file mode 100644
index 000000000..8eca29172
--- /dev/null
+++ b/reacnetgenerator/_mergeiso.py
@@ -0,0 +1,45 @@
+from tqdm import tqdm
+import numpy as np
+
+from .utils import bytestolist, listtobytes,  SharedRNGData
+
+
+class _mergeISO(SharedRNGData):
+    def __init__(self, rng):
+        SharedRNGData.__init__(
+            self, rng, ['miso', 'moleculetempfilename'], ['temp1it'])
+
+    def merge(self):
+        if self.miso > 0:
+            self._mergeISO()
+        self.returnkeys()
+
+    def _mergeISO(self):
+        with open(self.moleculetempfilename, 'rb') as ft:
+            items = ft.readlines()
+        items = [[items[i], items[i+1], items[i+2]]
+                 for i in range(0, len(items), 3)]
+        new_items = []
+        _oldbitem = b''
+        _oldbbond = b'0'
+        _oldindex = []
+        _oldfreq = 0
+        for _bitem, _bbond, _bindex in tqdm(sorted(items, key=lambda x: (x[0], x[1].split()[0])), desc='Merge isomers:'):
+            _index = bytestolist(_bindex)
+            _freq = len(_index)
+            if (_bitem == _oldbitem) and ((_bbond.split()[0] == _oldbbond.split()[0]) or (self.miso > 1)):
+                _oldindex = np.hstack((_oldindex, _index))
+            else:
+                if _oldbitem:
+                    new_items.append(
+                        [_oldbitem, _oldbbond, listtobytes(_oldindex)])
+                _oldbitem = _bitem
+                _oldindex = _index
+                if _freq > _oldfreq:
+                    _oldbbond = _bbond
+        new_items.append([_oldbitem, _oldbbond, listtobytes(_oldindex)])
+        new_items.sort(key=lambda x: len(x[0]))
+        self.temp1it = len(new_items)
+        with open(self.moleculetempfilename, 'wb') as ft:
+            for item in new_items:
+                [ft.write(i) for i in item]
diff --git a/reacnetgenerator/_path.py b/reacnetgenerator/_path.py
index b57a00857..a3afa97bb 100644
--- a/reacnetgenerator/_path.py
+++ b/reacnetgenerator/_path.py
@@ -19,6 +19,7 @@
 import itertools
 from abc import ABCMeta, abstractmethod
 from collections import Counter, defaultdict
+from re import S
 
 import networkx as nx
 import networkx.algorithms.isomorphism as iso
@@ -130,7 +131,11 @@ def convertSMILES(self, atoms, bonds):
 
     def _getatomsandbonds(self, line):
         atoms = np.array(bytestolist(line[0]), dtype=int)
-        bonds = bytestolist(line[1])
+        # bonds = bytestolist(line[1])
+        s = line[1].split()
+        pairs = bytestolist(s[0])
+        levels = bytestolist(s[1])
+        bonds = [[*pair, level] for pair, level in zip(pairs, levels)]
         return atoms, bonds
 
 
diff --git a/reacnetgenerator/commandline.py b/reacnetgenerator/commandline.py
index 35c1fa366..a310a3301 100644
--- a/reacnetgenerator/commandline.py
+++ b/reacnetgenerator/commandline.py
@@ -15,6 +15,8 @@ def _commandline():
     parser.add_argument(
         '--nohmm', help='Process trajectory without Hidden Markov Model (HMM)',
         action="store_true")
+    parser.add_argument(
+        '--miso', help='Merge the isomers',type=int, default=0)
     parser.add_argument(
         '--dump', help='Process the LAMMPS dump file', action="store_true")
     parser.add_argument(
@@ -44,6 +46,7 @@ def _commandline():
     import numpy as np
     ReacNetGenerator(
         inputfilename=args.inputfilename, atomname=args.atomname,
+        miso=args.miso,
         runHMM=not args.nohmm,
         inputfiletype=('lammpsdumpfile' if args.dump else args.type),
         nproc=args.nproc, selectatoms=args.selectatoms,
diff --git a/reacnetgenerator/dps.pyx b/reacnetgenerator/dps.pyx
index f4541416a..c4a0d395f 100644
--- a/reacnetgenerator/dps.pyx
+++ b/reacnetgenerator/dps.pyx
@@ -37,7 +37,8 @@ def dps(bonds, levels):
                 for b, l in zip(bonds[s], levels[s]):
                     b_c = b
                     if visited[b_c]==0:
-                        bond.append((s, b, l) if i < b else (b, s, l))
+                        # bond.append((s, b, l) if i < b else (b, s, l))
+                        bond.append((s, b, l) if s < b else (b, s, l))
                         st.push(b_c)
                 visited[s]=1
             mol.sort()
diff --git a/reacnetgenerator/reacnetgen.py b/reacnetgenerator/reacnetgen.py
index e6a9da126..4be61da70 100644
--- a/reacnetgenerator/reacnetgen.py
+++ b/reacnetgenerator/reacnetgen.py
@@ -53,6 +53,7 @@
 from . import __version__, __date__, __update__
 from ._detect import _Detect
 from ._draw import _DrawNetwork
+from ._mergeiso import _mergeISO 
 from ._hmmfilter import _HMMFilter
 from ._matrix import _GenerateMatrix
 from ._path import _CollectPaths
@@ -83,6 +84,11 @@ class ReacNetGenerator:
     runHMM: bool, optional, default: True
         Process trajectory with Hidden Markov Model (HMM) or not. If the user find too many species
         are filtered, they can turn off this option.
+    miso: int, optional, default: 0
+        Merge the isomers and the highest frequency is used as the representative. 0, off 
+        two available levels:
+        1, merge the isomers with same atoms and same bond-network but different bond levels;
+        2, merge the isomers with same atoms with different bond-network.
     pbc: bool, optional, default: True
         Use periodic boundary conditions (PBC) or not.
     cell: (3,3) array_like or (3,) array_like or (9,) array_like, optional, default: None
@@ -125,6 +131,7 @@ def __init__(self, **kwargs):
                             "nproc": cpu_count(), "pos": {}, "pbc": True, "split": 1, "n_searchspecies": 2,
                             "node_size": 200, "font_size": 6, "widthcoefficient": 1, "maxspecies": 20, "stepinterval": 1,
                             "nolabel": False,  "printfiltersignal": False, "showid": True, "runHMM": True, "SMILES": True,
+                            "miso": 0,
                             "getoriginfile": False, "needprintspecies": True, "urls": [],
                             "matrix_size":100,
                             }
@@ -179,6 +186,7 @@ def runanddraw(self, run=True, draw=True, report=True):
                 processthing.append(self.Status.DOWNLOAD)
             processthing.extend((
                 self.Status.DETECT,
+                self.Status.MISO,
                 self.Status.HMM,
                 self.Status.PATH,
                 self.Status.MATRIX,
@@ -201,6 +209,7 @@ def run(self):
             processthing.append(self.Status.DOWNLOAD)
         processthing.extend((
             self.Status.DETECT,
+            self.Status.MISO,
             self.Status.HMM,
             self.Status.PATH,
             self.Status.MATRIX,
@@ -233,6 +242,7 @@ class Status(Enum):
             - DOWNLOAD: Download trajectory from urls
             - DETECT: Read bond information and detect molecules
             - HMM: HMM filter
+            - MISO: Merge isomers
             - PATH: Indentify isomers and collect reaction paths
             - MATRIX: Reaction matrix generation
             - NETWORK: Draw reaction network
@@ -241,6 +251,7 @@ class Status(Enum):
 
         INIT = "Init"
         DETECT = "Read bond information and detect molecules"
+        MISO = "Merge isomers"
         HMM = "HMM filter"
         PATH = "Indentify isomers and collect reaction paths"
         MATRIX = "Reaction matrix generation"
@@ -264,6 +275,8 @@ def _process(self, steps):
         for i, runstep in enumerate(steps, 1):
             if runstep == self.Status.DETECT:
                 _Detect.gettype(self).detect()
+            elif runstep == self.Status.MISO:  
+                _mergeISO(self).merge()
             elif runstep == self.Status.HMM:
                 _HMMFilter(self).filter()
             elif runstep == self.Status.PATH: