"Terminal node has multiple features" error for spinal segments

[Kayvv]

There are hundreds of errors of the form:
"msg": "Terminal node has multiple features".
For example:

    {
        "type": "network",
        "network": "neural",
        "path": "ilxtr:sparc-nlp/mmset1/12",
        "entity": "UBERON:0001780",
        "features": [
            "S50_C1",
            "S50_C2",
            "S50_C3",
            "S50_C4",
            "S50_C5",
            "S50_C6",
            "S50_C7",
            "S50_C8",
            "S50_L1",
            "S50_L2",
            "S50_L3",
            "S50_L4",
            "S50_L5",
            "S50_L6",
            "S50_S1",
            "S50_S2",
            "S50_S3",
            "S50_S4",
            "S50_T1",
            "S50_T10",
            "S50_T11",
            "S50_T12",
            "S50_T13",
            "S50_T2",
            "S50_T3",
            "S50_T4",
            "S50_T5",
            "S50_T6",
            "S50_T7",
            "S50_T8",
            "S50_T9"
        ],
        "level": "error",
        "timestamp": "2025-07-03T03:12:09.675195Z",
        "msg": "Terminal node has multiple features"
    },

These features represent spinal cord segments spanning cervical, thoracic, lumbar, and sacral levels. It seems reasonable for them to share the same anatomical target: UBERON:0001780 (spinal nerve).
And the path "ilxtr:sparc-nlp/mmset1/12" rendered correctly on the flatmap.

• Should we allow multiple spinal segment features to map to the same UBERON ID? (I searched on SciCrunch but couldn’t find distinct UBERON terms for each spinal segment)
• Should the validator be adjusted to ignore or downgrade this error in such cases, assuming the path still renders correctly?
• Would it make sense to group the spinal segment features together and treat them as one unit, so the validator doesn’t complain? (I’m currently exploring this option, though I’m not sure whether it will affect how the path is rendered yet.)

[Kayvv]

I’ve tested two possible solutions to the “Terminal node has multiple features” error:

1. Add specific annotations + proxies to all features

I assigned each feature its own anatomical annotation (e.g., “First cervical nerve” instead of mapping all to “spinal nerve”) and added a proxy to UBERON:0001780 (spinal nerve) for each one.

However, this still resulted in the same error.

Add proxy to all spinal nerves:

    {
        "type": "network",
        "network": "neural",
        "path": "ilxtr:sparc-nlp/mmset1/12",
        "entity": "UBERON:0001780",
        "features": [
            "proxy_0_on_S50_C1",
            "proxy_0_on_S50_C2",
            "proxy_0_on_S50_C3",
            "proxy_0_on_S50_C4",
            "proxy_0_on_S50_C5",
            "proxy_0_on_S50_C6",
            "proxy_0_on_S50_C7",
            "proxy_0_on_S50_C8",
            "proxy_0_on_S50_L1",
            "proxy_0_on_S50_L2",
            "proxy_0_on_S50_L3",
            "proxy_0_on_S50_L4",
            "proxy_0_on_S50_L5",
            "proxy_0_on_S50_L6",
            "proxy_0_on_S50_S1",
            "proxy_0_on_S50_S2",
            "proxy_0_on_S50_S3",
            "proxy_0_on_S50_S4",
            "proxy_0_on_S50_T1",
            "proxy_0_on_S50_T10",
            "proxy_0_on_S50_T11",
            "proxy_0_on_S50_T12",
            "proxy_0_on_S50_T2",
            "proxy_0_on_S50_T3",
            "proxy_0_on_S50_T4",
            "proxy_0_on_S50_T5",
            "proxy_0_on_S50_T6",
            "proxy_0_on_S50_T7",
            "proxy_0_on_S50_T8",
            "proxy_0_on_S50_T9",
            "proxy_1_on_S50_L4",
            "proxy_1_on_S50_L6",
            "proxy_1_on_S50_T10"
        ],
        "level": "error",
        "timestamp": "2025-07-15T01:36:01.670665Z",
        "msg": "Terminal node has multiple features"
    },

Note: I wasn’t able to find UBERON or ILX terms for some of the segment-specific nerves (e.g., T13 or L6). This might be because humans do not typically have a 13th thoracic or 6th lumbar spinal nerve.

---

2. Add proxy to only one feature

In the second approach, I added the proxy only to a single segment (e.g., T1). This avoids triggering the error, but feels like a workaround rather than a proper solution. Also, the resulting flatmap rendering is less clear and visually less informative:

[Kayvv]

Since the path logic knows how to select the appropriate subset of spinal nerves, this situation doesn't seem like a true error. It may be more appropriate to treat it as a warning.