Merge branch 'main' of https://github.com/whitead/sanclone

.ruff.toml

@@ -0,0 +1,2 @@
+# Allow lines to be as long as 120 characters.
+line-length = 120

sanclone/agent/__init__.py

@@ -0,0 +1,3 @@
+from agent import SanCloneAgent
+
+__all__ = ["SanCloneAgent"]

sanclone/agent/agent.py

@@ -0,0 +1,41 @@
+from langchain.agents import AgentExecutor, ZeroShotAgent
+from langchain.agents.openai_functions_agent.base import OpenAIFunctionsAgent
+from langchain.chat_models import ChatOpenAI
+
+from ..tools import make_tools
+from .prompt import prompt_template
+
+
+class AgentType:
+    valid_models = {
+        "ReactAgent": ZeroShotAgent,
+        "OpenAIFunctionsAgent": OpenAIFunctionsAgent,
+    }
+
+    @classmethod
+    def get_agent(cls, model_name: str = "ReactAgent"):
+        return cls.valid_models[model_name]
+
+
+class SanCloneAgent:
+    def __init__(
+        self,
+        tools=None,
+        llm=None,
+        openai_api_key=None,
+        temp=0.1,
+        agent_type: str = "OpenAIFunctionsAgent",
+        verbose=True,
+    ):
+        llm = ChatOpenAI(temperature=0.0, model="gpt-4", client=None)
+
+        tools = make_tools(llm)
+        self.agent_instance = AgentExecutor.from_agent_and_tools(
+            tools=tools,
+            agent=AgentType.get_agent(agent_type).from_llm_and_tools(llm, tools),
+            return_intermediate_steps=True,
+            handle_parsing_errors=True,
+        )
+
+    def run(self, prompt: str):
+        return self.agent_instance.run(prompt_template.format(input=prompt))

sanclone/agent/prompt.py

@@ -0,0 +1,8 @@
+# flake8: noqa
+prompt_template = """
+You are an expert in molecular cloning.
+You have a set of tools at your disposal.
+Your task is to respond to the question or
+solve the problem to the best of your ability using the provided tools.
+Here is the question: {input}
+"""

sanclone/main.py

@@ -0,0 +1,37 @@
+# -*- coding: utf-8 -*-
+import click
+
+WELCOME = """
+Welcome to San Clone 👋  a molecular cloning agent 🧬.
+Give it an instruction like "Clone NADH Oxidase from Streptococcus pyogenes into pET16b"
+and press ✨ enter ✨
+"""
+
+
+@click.command()
+def main():
+    # check openai key
+    try:
+        from langchain.llms import OpenAI
+
+        OpenAI(model="babbage-002")
+    except Exception as e:
+        if "OPENAI_API_KEY" in str(e):
+            print("You need to set your OPENAI_API_KEY environment variable.")
+            print("You can get one from https://beta.openai.com/")
+            print("Then run the following command:")
+            print("export OPENAI_API_KEY=<your key>")
+            print("You can add this to your ~/.bashrc or ~/.bash_profile")
+            return
+    print(WELCOME)
+    while True:
+        instruction = input(">")
+        if instruction == "exit" or instruction == "quit" or instruction == "q":
+            print("Goodbye 👋")
+            break
+        else:
+            pass
+
+
+if __name__ == "__main__":
+    main()

sanclone/state.py

@@ -1,5 +1,51 @@
-from pydantic import BaseModel
+# -*- coding: utf-8 -*-
+"""get_seq_annotation.ipynb
 
+Automatically generated by Colaboratory.
+
+Original file is located at
+    https://colab.research.google.com/drive/1dJEOj6Jw3qOHsxcsP-W3Uj7mbKhvWnHi
+"""
+
+from Bio import Entrez
+from Bio import SeqIO
+
+
+class State:
+    def __init__(self):
+        self.vector = None
+        self.linear_insert = None
+        self.clone_seq = None
+
+    def store_vector(self, vector):
+        if not isinstance(vector, SeqIO.SeqRecord):
+                raise ValueError("Input 'vector' must be a SeqRecord object from Biopython's SeqIO.")
+        self.vector = vector
+
+    def store_linear_insert(self, linear_insert):
+        # Ensure the linear_insert is a SeqRecord object from Biopython
+        if not isinstance(linear_insert, SeqIO.SeqRecord):
+            raise ValueError("Input 'linear_insert' must be a SeqRecord object from Biopython's SeqIO.")
+        self.linear_insert = linear_insert
+
+    def store_clone_seq(self, clone_seq):
+        # Ensure the linear_insert is a SeqRecord object from Biopython
+        self.clone_seq = clone_seq
+
+    def retrieve_vector(self):
+        return self.vector
+
+    def retrieve_linear_insert(self):
+        return self.linear_insert
+
+    def retrieve_clone_seq(self):
+        return self.clone_seq
+    # seq to annotation
+
+
+def download_genbank_file(accession, filename):
+    Entrez.email = "[email protected]"  # Always provide your email address when using NCBI's services
+    with Entrez.efetch(db="nucleotide", id=accession, rettype="gb", retmode="text") as handle:
+        with open(filename, 'w') as outfile:
+            outfile.write(handle.read())
 
-class State(BaseModel):
-    pass

sanclone/tools/__init__.py

@@ -1,3 +1,4 @@
 from .echo import EchoTool
+from .maketools import make_tools
 
-__all__ = ["EchoTool"]
+__all__ = ["EchoTool", "make_tools"]

sanclone/tools/maketools.py

@@ -0,0 +1,11 @@
+from langchain import agents
+from langchain.llms.base import BaseLanguageModel
+
+
+def make_tools(llm: BaseLanguageModel):
+    # add human input tool
+    tools = agents.load_tools(["human"], llm)
+
+    # append tools here
+    tools += []
+    return tools

setup.py

@@ -17,7 +17,8 @@
     url="https://github.com/whitead/sanclone",
     license="MIT",
     packages=["sanclone", "sanclone.tools"],
-    install_requires=["langchain", "biopython"],
+    install_requires=["langchain", "biopython", "click"],
+    entry_points={"console_scripts": ["sanclone = sanclone.main:main"]},
     test_suite="tests",
     long_description=long_description,
     long_description_content_type="text/markdown",

tests/test_sanity.py

@@ -11,3 +11,28 @@ def test_echo_tool():
 
     tool = EchoTool(shared_state=State())
     assert tool.run("Hello") == "Hello"
+
+# def test_state_tool():
+#     from sanclone import State
+#     from sanclone.State import download_genbank_file
+#     accession_id_vector = "NC_005213"
+#     output_filename_vector = "NC_005213.gbk"
+#     accession_id_linear_insert = "NC_000932"
+#     output_filename_linear_insert = "NC_000932.gbk"
+#     download_genbank_file(accession_id_vector, output_filename_vector)
+#     download_genbank_file(accession_id_linear_insert, output_filename_linear_insert)
+
+#     for gb_record in SeqIO.parse(open(output_filename_linear_insert,"r"), "genbank") :
+#         # now do something with the record
+#         print ("Name %s, %i features" % (gb_record.name, len(gb_record.features)))
+
+#     vector_seq = list(SeqIO.parse(open(output_filename_vector,"r"), "genbank"))
+#     insert_seq = list(SeqIO.parse(open(output_filename_linear_insert,"r"), "genbank"))
+
+#     seq_anno = State(vector_seq[0])
+#     seq_anno.store_linear_insert(insert_seq[0])
+
+#     retrieved_vector = seq_anno.retrieve_vector()
+#     retrieved_insert = seq_anno.retrieve_linear_insert()
+#     print(retrieved_vector)
+#     print(retrieved_insert)