adding MR and LogP Prediction tasks

internbootcamp/bootcamp/ChemStructure2Property/InChI2MRBootCamp.py

@@ -0,0 +1,28 @@
+from rdkit import Chem
+from rdkit.Chem import Crippen
+from .InChI2logPBootCamp import InChI2logPbootcamp
+
+class InChI2MRBootCamp(InChI2logPbootcamp):
+
+    def prompt_func(self, InChI) -> str:
+
+        instruction = f"Given the InChI, determine the Molar Refractivity (MR) value of the material. The InChI is: {InChI}"
+        instruction_following = """Let's think step by step and output the final answer within \\boxed{}.The final answer should be one float number. For example "Final Answer: \\boxed{afloat}"."""
+        
+        prompt = instruction + '\n' + instruction_following
+        return prompt
+    
+        
+    @classmethod 
+    def _verify_correction(cls, solution, InChI)->bool:
+        """
+        Verify the correction of the solution.
+        """ 
+        mol = Chem.MolFromInchi(InChI)
+        true_MR = Crippen.MolMR(mol)
+        print(f"Comparing pred: {solution}, ground_truth: {true_MR}")
+        return abs(true_MR - float(solution)) <= 0.01  # maybe mse or mae better?
+        
+
+
+

internbootcamp/bootcamp/ChemStructure2Property/InChI2logPBootCamp.py

@@ -0,0 +1,56 @@
+from internbootcamp.bootcamp.base import Basebootcamp
+from internbootcamp.libs.chemStructure2Property.ChemStructureGenerator import InChIGenerator
+from .utils import last_boxed_only_string, remove_boxed
+from rdkit import Chem
+from rdkit.Chem import Crippen
+
+
+class InChI2logPbootcamp(Basebootcamp):
+    def __init__(self, num_numbers=4, max_atoms=15, min_atoms=3, elements=None, seed=None):
+        # super.__init__()
+        self.num_numbers = num_numbers
+        self.InChIGenerator = InChIGenerator(max_atoms=max_atoms, min_atoms=min_atoms, elements=elements, seed=seed)
+        
+    def case_generator(self) -> str:
+        """
+        生成一组数字和目标值。
+        """
+        return self.InChIGenerator.generate_n_valid_inchi(1)[0]
+
+    def prompt_func(self, InChI) -> str:
+
+        instruction = f"Given the InChI, determine the lipophilicity (logP) value of the material. The InChI is: {InChI}"
+        instruction_following = """Let's think step by step and output the final answer within \\boxed{}.The final answer should be one float number. For example "Final Answer: \\boxed{afloat}"."""
+        
+        prompt = instruction + '\n' + instruction_following
+        return prompt
+    
+    @staticmethod
+    def extract_output(output):
+        """
+        Extract the output from the solution.
+        
+        Args:
+            output: Model output to be processed.
+        
+        Returns:
+            The processed output.
+        """
+        output = last_boxed_only_string(output)
+        if output is None:
+            return None
+        return remove_boxed(output)
+        
+    @classmethod 
+    def _verify_correction(cls, solution, InChI)->bool:
+        """
+        Verify the correction of the solution.
+        """ 
+        mol = Chem.MolFromInchi(InChI)
+        true_logp = Crippen.MolLogP(mol)
+        print(f"Comparing pred: {solution}, ground_truth: {true_logp}")
+        return abs(true_logp - float(solution)) <= 0.01  # maybe mse or mae better?
+        
+
+
+

internbootcamp/bootcamp/ChemStructure2Property/SMILES2MRBootCamp.py

@@ -0,0 +1,34 @@
+from internbootcamp.bootcamp.base import Basebootcamp
+from internbootcamp.libs.chemStructure2Property.ChemStructureGenerator import SMILESGenerator
+from .utils import last_boxed_only_string, remove_boxed
+from rdkit import Chem
+from rdkit.Chem import Crippen
+
+from .SMILES2logPBootCamp import SMILES2logPBootCamp
+
+class SMILES2MRBootCamp(SMILES2logPBootCamp):
+
+
+
+    def prompt_func(self,  SMILES) -> str:
+
+        instruction = f"Given the  SMILES, determine the Molar Refractivity (MR) value of the material. The  SMILES is: {SMILES}"
+        instruction_following = """Let's think step by step and output the final answer within \\boxed{}.The final answer should be one float number. For example "Final Answer: \\boxed{afloat}"."""
+        
+        prompt = instruction + '\n' + instruction_following
+        return prompt
+
+        
+    @classmethod 
+    def _verify_correction(cls, solution, SMILES)->bool:
+        """
+        Verify the correction of the solution.
+        """ 
+        mol = Chem.MolFromSmiles(SMILES)
+        true_MR = Crippen.MolMR(mol)
+        print(f"Comparing pred: {solution}, ground_truth: {true_MR}")
+        return abs(true_MR - float(solution)) <= 0.01  # maybe mse or mae better?
+        
+
+
+

internbootcamp/bootcamp/ChemStructure2Property/SMILES2logPBootCamp.py

@@ -0,0 +1,43 @@
+from internbootcamp.bootcamp.base import Basebootcamp
+from internbootcamp.libs.chemStructure2Property.ChemStructureGenerator import SMILESGenerator
+from .utils import last_boxed_only_string, remove_boxed
+from rdkit import Chem
+from rdkit.Chem import Crippen
+
+from .InChI2logPBootCamp import InChI2logPbootcamp
+
+class SMILES2logPBootCamp(InChI2logPbootcamp):
+    def __init__(self, num_numbers=4, min_len=5, max_len=25, 
+                 seed=None):
+        # super.__init__()
+        self.num_numbers = num_numbers
+        self.SMILESGenerator = SMILESGenerator(min_len=5, max_len=25, seed=None)
+        
+    def case_generator(self) -> str:
+        """
+        生成一组数字和目标值。
+        """
+        return self.SMILESGenerator.generate_n_valid_smiles(1)[0]
+
+    def prompt_func(self,  SMILES) -> str:
+
+        instruction = f"Given the  SMILES, determine the lipophilicity (logP) value of the material. The  SMILES is: {SMILES}"
+        instruction_following = """Let's think step by step and output the final answer within \\boxed{}.The final answer should be one float number. For example "Final Answer: \\boxed{afloat}"."""
+        
+        prompt = instruction + '\n' + instruction_following
+        return prompt
+
+        
+    @classmethod 
+    def _verify_correction(cls, solution, SMILES)->bool:
+        """
+        Verify the correction of the solution.
+        """ 
+        mol = Chem.MolFromSmiles(SMILES)
+        true_logp = Crippen.MolLogP(mol)
+        print(f"Comparing pred: {solution}, ground_truth: {true_logp}")
+        return abs(true_logp - float(solution)) <= 0.01  # maybe mse or mae better?
+        
+
+
+

internbootcamp/bootcamp/ChemStructure2Property/utils.py

@@ -0,0 +1,43 @@
+
+def remove_boxed(s):
+    if "\\boxed " in s:
+        left = "\\boxed "
+        assert s[:len(left)] == left
+        return s[len(left):]
+
+    left = "\\boxed{"
+
+    assert s[:len(left)] == left
+    assert s[-1] == "}"
+
+    return s[len(left):-1]
+
+
+def last_boxed_only_string(string):
+    idx = string.rfind("\\boxed")
+    if "\\boxed " in string:
+        return "\\boxed " + string.split("\\boxed ")[-1].split("$")[0]
+    if idx < 0:
+        idx = string.rfind("\\fbox")
+        if idx < 0:
+            return None
+
+    i = idx
+    right_brace_idx = None
+    num_left_braces_open = 0
+    while i < len(string):
+        if string[i] == "{":
+            num_left_braces_open += 1
+        if string[i] == "}":
+            num_left_braces_open -= 1
+            if num_left_braces_open == 0:
+                right_brace_idx = i
+                break
+        i += 1
+
+    if right_brace_idx is None:
+        retval = None
+    else:
+        retval = string[idx:right_brace_idx + 1]
+
+    return retval