aurelian 0.3.2__py3-none-any.whl

aurelian/agents/gocam/documents/pandoc_md/Signaling_receptor_activity_annotation_guidelines.md

@@ -0,0 +1,334 @@
+Receptor activation by a ligand is represented differently in GO-CAM
+depending on whether the ligand is (i) a protein (i. e., encoded by a
+gene) or (ii) a small molecule. 
+
+Pathway Editor
+==============
+
+Protein ligand-activated signaling receptor
+-------------------------------------------
+
+### Ligand
+
+The activity unit for a [ligand of a signaling receptor]{.underline} is:
+
+-   **MF**: a ligand \'[enables]{.underline}\' receptor ligand activity
+    > ([[GO:0048018]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0048018))
+    > or a child
+
+-   **Context:**
+
+    -   The relation between a ligand and its target receptor is
+        > *\'*[has input\']{.underline}
+
+    -   **BP** \'[part of\']{.underline} the process in which the ligand
+        > participates, usually a child of signal transduction
+        > ([[GO:0007165]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0007165))
+
+    -   **CC:**
+
+        -   extracellular ligands: \'[occurs in]{.underline}\'
+            > extracellular space
+            > ([[GO:0005615]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0005886))
+
+        -   membrane-bound ligands: \'[occurs in]{.underline}\' plasma
+            > membrane
+            > ([[GO:0005886]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0005886))
+
+    -   The causal relation between the **ligand activity** and the
+        > **receptor activity** is\
+        > \'[directly positively regulates]{.underline}\'.
+
+### Signaling receptor
+
+The activity unit for a [signaling receptor]{.underline} is:
+
+-   **MF:** \'[enables]{.underline}\' signaling receptor activity
+    > ([[GO:0038023]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0038023)*)*
+    > or a child*.* 
+
+-   **Context:**
+
+    -   The input (target) of the receptor is the **effector protein it
+        > regulates**, for example a molecular adaptor, captured with
+        > the *\'*[has input\']{.underline} relation.
+
+    -   **Note that the input (target) of the receptor is NOT its
+        > ligand**
+
+-   **BP** \'[part of\']{.underline} the same BP/signal transduction
+    > pathway as the ligand
+
+-   **CC:** transmembrane receptors: \'[occurs in]{.underline}\' plasma
+    > membrane
+    > ([[GO:0005886]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0005886))
+
+    -   The causal relation between the MF of the **receptor** and the
+        > MF of its **target** is \'[directly positively
+        > regulates\']{.underline}*.*
+
+**Example: [[Insulin signaling
+model]{.underline}](http://noctua.geneontology.org/workbench/noctua-visual-pathway-editor/?model_id=gomodel%3A6482692800000931)**
+
+![](media/image5.png){width="6.267716535433071in"
+height="1.3888888888888888in"}
+
+ 
+
+Small molecule-activated signaling receptor activity
+----------------------------------------------------
+
+-   Since small molecules are not annotated in GO, a small molecule
+    > ligand does not have a molecular function. Instead, the **ligand**
+    > and the **receptor activity** are linked by the causal relation
+    > \'[is small molecule activator\']{.underline}*.*
+
+-   The receptor\'s function, input, and contextual relations are the
+    > same as for protein ligand-activated receptors. 
+
+-   Note that it is also possible to annotate an inhibitory ligand using
+    > the relation \'[is small molecule inhibitor\']{.underline}*.*
+
+-   BP and CC annotations are the same as for protein ligand-activated
+    > receptor activity.
+
+**Example: [[Activation of a GPCR by
+succinate]{.underline}](http://noctua.geneontology.org/workbench/noctua-visual-pathway-editor/?model_id=gomodel%3A648d0dc100000022)**
+
+![](media/image1.png){width="5.526042213473316in"
+height="2.5371741032370956in"}
+
+ 
+
+Special cases
+-------------
+
+### 
+
+### Receptor with coreceptor 
+
+This is common in immune receptors. The typical sequence of events is
+that the ligand binds the signaling receptor, which signals to the
+co-receptor to activate its downstream effector (such as a protein
+kinase).
+
+**Example: [[IL2 signaling pathway
+(Human)]{.underline}](http://noctua.geneontology.org/workbench/noctua-visual-pathway-editor/?model_id=gomodel:6205c24300001663)**
+
+In this case, the receptor is activated by an information
+biomacromolecule (i.e. a protein). Interleukin-2
+([[IL2]{.underline}](https://www.uniprot.org/uniprot/P60568)), a
+cytokine, activates its receptor, interleukin-2 receptor A
+([[IL2RA]{.underline}](https://www.uniprot.org/uniprot/P01589)). 
+
+IL2RA directly positively regulates (activates)
+[[IL2RB]{.underline}](https://www.uniprot.org/uniprot/P14784), which
+phosphorylates and positively regulates
+[[JAK1]{.underline}](https://www.uniprot.org/uniprotkb/P23458).
+
+The activity unit for a [signaling coreceptor]{.underline} is:
+
+-   **MF:** coreceptor activity
+    > ([[GO:0015026]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0015026)*)*
+
+-   **Context:**
+
+    -   The relation between the receptor and its input (target) is the
+        > **signaling coreceptor** it regulates, captured with the
+        > *\'*[has input]{.underline}\' relation
+
+    -   **Note that the input (target) of the coreceptor is NOT the
+        > ligand**
+
+-   **BP** \'[part of]{.underline}\' the BP in which the signaling
+    > receptor is involved, usually a child of signal transduction
+    > ([[GO:0007165]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0007165))
+
+-   **CC:** transmembrane receptors: \'[occurs in]{.underline}\' plasma
+    > membrane
+    > ([[GO:0005886]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0005886))
+
+    -   The causal relation between the MF of the **receptor** and the
+        > MF of its **coreceptor** is \'[directly positively
+        > regulates]{.underline}\'*.*
+
+![Diagram Description automatically generated with medium
+confidence](media/image2.png){width="3.3854166666666665in"
+height="1.1354166666666667in"}
+
+Note: 
+
+-   cytokine activity is\_a receptor ligand activity
+
+-   interleukin-2 receptor activity is\_a signaling receptor activity
+
+### 
+
+### 
+
+ 
+
+Form Editor
+===========
+
+Protein ligand-activated signaling receptor
+-------------------------------------------
+
+### Ligand activity
+
+The activity unit for a [ligand of a signaling receptor]{.underline} is:
+
+-   **MF**: a ligand \'[enables]{.underline}\' receptor ligand activity
+    > ([[GO:0048018]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0048018))
+    > or a child
+
+-   **Context:**
+
+    -   The relation between a ligand and its target receptor is
+        > *\'*[has input\']{.underline}
+
+    -   **BP** \'[part of]{.underline}\' the BP in which this ligand is
+        > involved (usually a child of signal transduction
+        > ([[GO:0007165]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0007165)))
+
+    -   **CC:**
+
+        -   extracellular ligands: \'[occurs in]{.underline}\'
+            > extracellular space
+            > ([[GO:0005615]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0005886))
+
+        -   membrane-bound ligands: \'[occurs in]{.underline}\' plasma
+            > membrane
+            > ([[GO:0005886]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0005886))
+
+-   The causal relation between the ***ligand* *activity*** and the
+    > ***receptor activity*** is\
+    > \'[directly positively regulates]{.underline}\'.
+
+**Example: [[Insulin signaling
+model]{.underline}](http://noctua.geneontology.org/workbench/noctua-form/?model_id=gomodel%3A6482692800000931)**
+
+![](media/image3.png){width="6.267716535433071in"
+height="2.736111111111111in"}
+
+### Signaling receptor activity
+
+The activity unit for a [signaling receptor]{.underline} is:
+
+-   **MF:** The receptor \'[enables]{.underline}\' signaling receptor
+    > activity
+    > ([[GO:0038023]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0038023)*)*
+    > or a child*.* 
+
+-   **Context:**
+
+    -   The input (target) of the receptor is the **effector protein it
+        > regulates**, for example a molecular adaptor, captured with
+        > the *\'*[has input\']{.underline} relation
+
+    -   **Note that the input (target) of the receptor is NOT its
+        > ligand**
+
+-   **BP** in which the signaling receptor is involved, usually a child
+    > of signal transduction
+    > ([[GO:0007165]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0007165))
+
+-   **CC:** \'[occurs in]{.underline}\' a child of cellular anatomical
+    > entity (GO:0110165). For transmembrane receptors, annotate to
+    > plasma membrane
+    > ([[GO:0005886]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0005886))
+
+**Example: [[Insulin signaling
+model]{.underline}](http://noctua.geneontology.org/workbench/noctua-form/?model_id=gomodel%3A6482692800000931)**
+
+![](media/image7.png){width="6.267716535433071in" height="3.125in"}
+
+-   The causal relation between the MF of the **receptor** and the MF of
+    > its **target** is \'[directly positively
+    > regulates\']{.underline}*.*
+
+ 
+
+Small molecule-activated signaling receptor activity
+----------------------------------------------------
+
+-   In the Form, small molecules are not captured, so the MF for the
+    > receptor. Therefore, the MF is \'[enables]{.underline}\' signaling
+    > receptor activity
+    > ([[GO:0038023]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0038023)*)*
+    > or a child*.* 
+
+-   BP and CC annotations are the same as for protein ligand-activated
+    > receptor activity.
+
+![](media/image4.png){width="6.267716535433071in"
+height="2.7916666666666665in"}
+
+Receptor with coreceptor 
+-------------------------
+
+### 
+
+**Example: [[IL2 signaling pathway
+(Human)]{.underline}](http://noctua.geneontology.org/workbench/noctua-form/?model_id=gomodel%3A6205c24300001663)**
+
+In this case, the receptor, interleukin-2 receptor A
+([[IL2RA]{.underline}](https://www.uniprot.org/uniprot/P01589)), is
+activated by a protein ligand, interleukin-2
+([[IL2]{.underline}](https://www.uniprot.org/uniprot/P60568)). IL2RA
+directly positively regulates (activates) the coreceptor
+[[IL2RB]{.underline}](https://www.uniprot.org/uniprot/P14784), which
+phosphorylates and positively regulates
+[[JAK1]{.underline}](https://www.uniprot.org/uniprotkb/P23458).
+
+The activity unit for a [signaling coreceptor]{.underline} is:
+
+-   **MF:** The coreceptor \'[enables]{.underline}\' coreceptor activity
+    > ([[GO:0015026]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0015026)*)*
+    > or a child*.* 
+
+-   **Context:**
+
+    -   The input (target) of the coreceptor is the **signaling
+        > receptor** it regulates, captured with the *\'*[has
+        > input]{.underline}\' relation
+
+    -   **Note that the input (target) of the coreceptor is NOT the
+        > ligand**
+
+-   **BP** \'[part of]{.underline}\' the BP in which the signaling
+    > receptor is involved, usually a child of signal transduction
+    > ([[GO:0007165]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0007165))
+
+-   **CC:** transmembrane receptors: \'[occurs in]{.underline}\' plasma
+    > membrane
+    > ([[GO:0005886]{.underline}](https://www.ebi.ac.uk/QuickGO/term/GO:0005886))
+
+![](media/image6.png){width="6.267716535433071in"
+height="2.3194444444444446in"}
+
+Note: 
+
+-   cytokine activity is\_a receptor ligand activity
+
+-   interleukin-2 receptor activity is\_a signaling receptor activity
+
+ 
+
+Differences between GO-CAM and standard annotation of a signaling receptor and its ligand
+-----------------------------------------------------------------------------------------
+
+In standard annotation (captured with the Noctua Form or Protein2GO),
+relations between molecular functions are not captured, so there is no
+relation between the ligand activity and the signaling receptor
+activity, nor is there a relation between the signaling receptor
+activity and the activity of its target. Likewise for receptor activity
+and coreceptor activity, there is no relation between these activities
+captured in standard annotations.
+
+Review information
+==================
+
+Review date: 2023-07-20
+
+Reviewed by: Cristina Casals, Pascale Gaudet, Patrick Masson

aurelian/agents/gocam/gocam_agent.py

@@ -0,0 +1,240 @@
+"""
+Agent for working with GO-CAMs (Gene Ontology Causal Activity Models).
+"""
+from enum import Enum
+from typing import List, Optional, Dict
+
+from pydantic import BaseModel, Field
+
+from aurelian.agents.gocam.gocam_config import GOCAMDependencies
+from aurelian.agents.gocam.gocam_tools import (
+    search_gocams,
+    lookup_gocam,
+    lookup_uniprot_entry,
+    all_documents,
+    fetch_document,
+    lookup_gocam_local,
+)
+from aurelian.agents.literature.literature_tools import (
+    lookup_pmid as literature_lookup_pmid,
+    search_literature_web,
+    retrieve_literature_page
+)
+from aurelian.agents.filesystem.filesystem_tools import inspect_file, list_files
+from pydantic_ai import Agent, Tool, RunContext
+
+class RecommendationStrength(str, Enum):
+    """
+    Enum for recommendation strength.
+    """
+    MAJOR = "major"
+    MODERATE = "moderate"
+    MINOR = "minor"
+
+class Recommendation(BaseModel):
+    """
+    Recommendation for improving a GO-CAM model.
+    """
+    text: str
+    strength: RecommendationStrength
+
+    class Config:
+        schema_extra = {
+            "example": {
+                "text": "Improve the clarity of the model diagram.",
+                "strength": RecommendationStrength.MAJOR
+            }
+        }
+
+class GOCamReviewSummary(BaseModel):
+    """
+    Summary of the review of a GO-CAM model with rubric scores.
+    """
+    model_id: str
+    model_title: str
+    taxa: Optional[List[str]] = Field(description="List of taxa (formal scientific names)")
+    genes: Optional[List[str]] = Field(description="List of genes in the model")
+    model_description: str = Field(description="Description of the model")
+    review_summary: str = Field(description="Brief summary of the review")
+    best_practice_score: Optional[int] = Field(None, description="Score 1-5 for adherence to GO-CAM best practices", ge=1, le=5)
+    specific_best_practice_scores: Optional[Dict[str, int]] = Field(None, description="Key-value set of scores for each relevant best practice. Each key is the name of a specific best practice (use the document name), and the value is the score 1-5. Only for relevant best practices")
+    biological_content_score: Optional[int] = Field(None, description="Score 1-5 for accuracy of biological content against literature", ge=1, le=5)
+    publication_consistency_score: Optional[int] = Field(None, description="Score 1-5 for consistency with publications", ge=1, le=5)
+    uniprot_consistency_score: Optional[int] = Field(None, description="Score 1-5 for consistency with uniprot information", ge=1, le=5)
+    causal_connections_score: Optional[int] = Field(None, description="Score 1-5 for logical connection of activities in pathway flow", ge=1, le=5)
+    simplicity_score: Optional[int] = Field(None, description="Score 1-5 for model parsimony and human understandability", ge=1, le=5)
+    completeness_score: Optional[int] = Field(None, description="Score 1-5 for model completeness", ge=1, le=5)
+    overall_score: Optional[int] = Field(None, description="Overall score 1-5 for the model", ge=1, le=5)
+    recommendations: Optional[List[Recommendation]] = Field(None, description="Key recommendations for improvement, highest priority first")
+
+
+SYSTEM_CURIES = """
+When providing results in markdown, you should generally include CURIEs/IDs, and you can 
+hyperlink these as https://bioregistry.io/{curie}. Note that GO-CAM IDs should be hyperlinked 
+as https://bioregistry.io/go.model:{uuid}."""
+
+PREDICATES_INFO = """
+The following predicates are used for causal associations:
+
+- RO:0002413 *provides input for*
+- RO:0002629 *directly positively regulates*
+- RO:0002630 *directly negatively regulates*
+- RO:0002304 *causally upstream of, positive effect*
+- RO:0002305 *causally upstream of, negative effect*
+- RO:0002307 *indirectly positively regulates*
+- RO:0002308 *indirectly negatively regulates*
+
+"""
+
+SYSTEM = f"""
+You are an expert molecular biologist with access to the GO-CAM database.
+
+GO-CAMs (Gene Ontology Causal Activity Models) are standardized models that represent 
+biological processes and pathways, including gene functions and interactions.
+
+You can help with:
+- Searching for GO-CAM models by pathway, gene, or complex queries
+- Looking up specific GO-CAM models by ID
+- Finding information about proteins via UniProt
+- Analyzing and comparing biological pathways
+- Retrieving literature related to GO-CAMs via PubMed
+- Retrieving GO-CAM annotation best practice documents
+- Validating GO-CAM model structure against the schema
+
+You can provide information on gene functions, pathways, and models. When giving your response, 
+stick to communicating the information provided in the response. You may extemporize and fill 
+in gaps with your own knowledge, but always be clear about what information came from the call 
+vs your own knowledge.
+
+{PREDICATES_INFO}
+
+{SYSTEM_CURIES}
+"""
+
+core_tools = [
+    Tool(search_gocams),
+    Tool(lookup_gocam),
+    Tool(lookup_uniprot_entry),
+    Tool(literature_lookup_pmid,
+         description="""Lookup the text of a PubMed article by its PMID.
+
+                        Note that assertions in GO-CAMs may reference PMIDs, so this tool
+                        is useful for validating assertions. A common task is to align
+                        the text of a PMID with the text of an assertion, or extracting text
+                        snippets from the publication that support the assertion."""),
+    Tool(search_literature_web),
+    Tool(retrieve_literature_page),
+    Tool(fetch_document),
+    Tool(inspect_file),
+    Tool(list_files),
+]
+
+
+gocam_agent = Agent(
+    model="openai:gpt-4o",
+    deps_type=GOCAMDependencies,
+    system_prompt=SYSTEM,
+    tools=core_tools,
+)
+
+def get_documents_for_prompt() -> str:
+    """
+    Get the documents for the system prompt.
+
+    Returns:
+        A string containing the list of available GO-CAM documents
+    """
+    meta = all_documents()
+    if not meta["documents"]:
+        return "\nNo GO-CAM best practice documents are available."
+
+    docs_text = "\n\nThe following GO-CAM best practice documents are available:\n"
+    docs_text += "\n".join([f"- {d['title']}" for d in meta["documents"]])
+    docs_text += "\n\nYou can use the `fetch_document` tool to retrieve the content of any of these documents."
+    docs_text += "\nWhen asked any question about GO-CAM curation practice, be sure to ALWAYS"
+    docs_text += " check the relevant document for the most up-to-date information.\n"
+    docs_text += "Some of these docs refer to particular exemplar models. these can be retrieved with the `lookup_gocam` tool."
+    return docs_text
+
+@gocam_agent.system_prompt
+def add_documents(ctx: RunContext[GOCAMDependencies]) -> str:
+    """
+    Add available GO-CAM documents to the system prompt.
+    
+    Args:
+        ctx: The run context
+        
+    Returns:
+        A string containing the list of available GO-CAM documents
+    """
+    return get_documents_for_prompt()
+
+
+REVIEWER_SYSTEM = f"""
+You are a GO-CAM curator in charge of reviewing proposed GO-CAMs.
+
+GO-CAMs (Gene Ontology Causal Activity Models) are standardized models that represent 
+biological processes and pathways, including gene functions and interactions.
+
+Your job is to examine proposed models, and to perform review and QC, including:
+
+- checking that the model is consistent with all relevant GO-CAM best practices (see `fetch_document`)
+- ensuring the biological content of the model is consistent with the literature and textbook knowledge (see `literature_lookup_pmid`)
+- everything is consistent with what is known about that protein (see `lookup_uniprot_entry`)
+- activities in the model are connected in a way that is consistent with the activity flow in the pathway
+- the model is parsimonious and easy for a human to understand
+
+For each review, you should score the model on a scale of 1-5 (5 being best) for each of these criteria:
+1. Best Practices: adherence to GO-CAM best practices
+2. Biological Content: accuracy of biological content against literature
+3. Protein Consistency: consistency with known protein functions and interactions
+4. Causal Connections: logical connection of activities in pathway flow
+5. Parsimony: model simplicity and human understandability
+
+Provide an overall score (1-5) and key recommendations for improvement at the end of your review.
+If any criteria are not applicable, mark them as N/A.
+
+{PREDICATES_INFO}
+
+{SYSTEM_CURIES}
+"""
+
+gocam_reviewer_agent = Agent(
+    model="anthropic:claude-3-7-sonnet-latest",
+    deps_type=GOCAMDependencies,
+    system_prompt=REVIEWER_SYSTEM,
+    tools=[
+        *core_tools,
+        Tool(lookup_gocam_local),
+        #Tool(validate_gocam_model),
+    ],
+)
+
+
+@gocam_reviewer_agent.system_prompt
+def add_documents(ctx: RunContext[GOCAMDependencies]) -> str:
+    """
+    Add available GO-CAM documents to the system prompt.
+
+    Args:
+        ctx: The run context
+
+    Returns:
+        A string containing the list of available GO-CAM documents
+    """
+    return get_documents_for_prompt()
+
+
+REVIEW_SUMMARIZER_SYSTEM = """Your job is to summarize the existing review of a GO-CAM model, completing
+the rubric for all of the criteria.
+"""
+
+gocam_review_summarizer_agent = Agent(
+    model="anthropic:claude-3-7-sonnet-latest",
+    deps_type=GOCAMDependencies,
+    system_prompt=REVIEW_SUMMARIZER_SYSTEM,
+    tools=[
+        #Tool(lookup_gocam_local),
+    ],
+    result_type=GOCamReviewSummary,
+)

aurelian/agents/gocam/gocam_config.py

@@ -0,0 +1,85 @@
+"""
+Configuration classes for the GOCAM agent.
+"""
+from dataclasses import dataclass, field
+import os
+from typing import Optional
+
+from bioservices import UniProt
+from linkml_store import Client
+from linkml_store.api import Collection
+
+from aurelian.dependencies.workdir import HasWorkdir, WorkDir
+
+# Default database connection settings
+HANDLE = "mongodb://localhost:27017/gocams"
+DB_NAME = "gocams"
+COLLECTION_NAME = "main"
+
+# Initialize UniProt service
+uniprot_service = UniProt(verbose=False)
+
+
+@dataclass
+class GOCAMDependencies(HasWorkdir):
+    """
+    Configuration for the GOCAM agent.
+    """
+    max_results: int = field(default=10)
+    db_path: str = field(default=HANDLE)
+    db_name: str = field(default=DB_NAME)
+    collection_name: str = field(default=COLLECTION_NAME)
+    _collection: Optional[Collection] = None
+    
+    def __post_init__(self):
+        """Initialize the config with default values."""
+        # Initialize workdir if not provided
+        if self.workdir is None:
+            self.workdir = WorkDir()
+
+    @property
+    def collection(self) -> Collection:
+        """
+        Get the GOCAM collection, initializing the connection if needed.
+        
+        Returns:
+            Collection: The GOCAM collection
+        """
+        if self._collection is None:
+            client = Client()
+            client.attach_database(self.db_path, alias=self.db_name)
+            db = client.databases[self.db_name]
+            self._collection = db.get_collection(self.collection_name)
+        return self._collection
+    
+    def get_uniprot_service(self) -> UniProt:
+        """
+        Get the UniProt service for protein lookups.
+        
+        Returns:
+            UniProt: The UniProt service
+        """
+        return uniprot_service
+
+
+def get_config() -> GOCAMDependencies:
+    """
+    Get the GOCAM agent configuration from environment variables or defaults.
+    
+    Returns:
+        GOCAMDependencies: The GOCAM dependencies
+    """
+    workdir_path = os.environ.get("AURELIAN_WORKDIR", None)
+    workdir = WorkDir(location=workdir_path) if workdir_path else None
+    
+    # Get any environment-specific settings
+    db_path = os.environ.get("GOCAM_DB_PATH", HANDLE)
+    db_name = os.environ.get("GOCAM_DB_NAME", DB_NAME)
+    collection_name = os.environ.get("GOCAM_COLLECTION", COLLECTION_NAME)
+    
+    return GOCAMDependencies(
+        workdir=workdir,
+        db_path=db_path,
+        db_name=db_name,
+        collection_name=collection_name
+    )