/go-try

Create GO-CAM annotations from natural language descriptions.

Usage

/go-try <description of the annotation>

Examples

/go-try annotate that human TP53 has protein kinase binding activity from PMID:20154695

/go-try mouse Brca1 has DNA repair activity in the nucleus, evidenced by IDA from PMID:12345678

/go-try create a model for Drosophila notch signaling with experimental evidence

What It Does

1. Extracts information from your description — gene product, organism, GO terms, evidence code, and reference
2. Asks clarifying questions if anything is missing (one round only)
3. Resolves identifiers by searching UniProt, GO, and other ontology services
4. Builds a GO-CAM model using the gocam-py library
5. Saves to output/ as a JSON file
6. Explains the output — what each field means and why specific terms were chosen

What You Need to Provide

Field	Required	Example
Gene product	Yes	"TP53", "Brca1", "EGFR"
Organism	Yes	"human", "mouse", "Drosophila"
Molecular function	Yes	"protein kinase binding", "DNA repair"
Evidence code	Yes	"IDA", "IMP", "from experiment"
Reference	Yes	"PMID:20154695"
Biological process	No	"apoptotic process"
Cellular component	No	"nucleus", "cytoplasm"

Natural Language Works

You don't need to use exact ontology terms. Say "protein binding" and Claude will find the right GO term. Say "from a paper" and Claude will ask for the PMID.

Output

Files are saved to output/ with the naming pattern:

output/gocam-YYYYMMDD-HHMMSS.json

Next Steps

After creating an annotation:

• /go-validate — Check it against the schema
• /go-check — Run quality checks
• /go-pr — Submit for review