feat: Thermo Fisher Scientific NanoDrop 8000 - add unread data to asm (#1094)

<img width="1805" height="162" alt="image"
src="https://github.com/user-attachments/assets/567064c5-c482-4c2d-b2f6-de45a7809d0b"
/>

Author: joshua-benchling

src/allotropy/allotrope/converter.py

@@ -112,6 +112,7 @@
     "=": "_EQUALS_",
     "@": "_AT_",
     "'": "_QUOTE_",
+    "*": "_ASTERISK_",
     ",": "_COMMA_",
     "&": "_AMPERSAND_",
     # NOTE: this MUST be at the end, or it will break other key replacements.

src/allotropy/allotrope/models/shared/definitions/custom.py

@@ -54,6 +54,7 @@
     PartsPerBillion,
     PartsPerMillion,
     Percent,
+    PerMolarPerCentimeter,
     PerMolarPerSecond,
     PerSecond,
     PH,
@@ -610,6 +611,18 @@ class TNullableQuantityValuePartsPerMillion(PartsPerMillion, TNullableQuantityVa
     pass
 
 
+@dataclass(frozen=True, kw_only=True)
+class TQuantityValuePerMolarPerCentimeter(PerMolarPerCentimeter, TQuantityValue):
+    pass
+
+
+@dataclass(frozen=True, kw_only=True)
+class TNullableQuantityValuePerMolarPerCentimeter(
+    PerMolarPerCentimeter, TNullableQuantityValue
+):
+    pass
+
+
 @dataclass(frozen=True, kw_only=True)
 class TQuantityValuePerMolarPerSecond(PerMolarPerSecond, TQuantityValue):
     pass

src/allotropy/allotrope/models/shared/definitions/units.py

@@ -249,6 +249,11 @@ class PartsPerMillion(HasUnit):
     unit: str = "ppm"
 
 
+@dataclass(frozen=True, kw_only=True)
+class PerMolarPerCentimeter(HasUnit):
+    unit: str = "M-1cm-1"
+
+
 @dataclass(frozen=True, kw_only=True)
 class PerMolarPerSecond(HasUnit):
     unit: str = "M-1s-1"

src/allotropy/allotrope/schema_mappers/adm/spectrophotometry/benchling/_2023/_12/spectrophotometry.py

@@ -29,9 +29,12 @@
     UltravioletAbsorbanceSpectrumDetectionMeasurementDocumentItems,
 )
 from allotropy.allotrope.models.shared.definitions.custom import (
+    TQuantityValueKiloDalton,
     TQuantityValueMicroliter,
     TQuantityValueMilliAbsorbanceUnit,
+    TQuantityValueNanogramPerMicroliter,
     TQuantityValueNanometer,
+    TQuantityValuePerMolarPerCentimeter,
     TQuantityValueRelativeFluorescenceUnit,
     TQuantityValueUnitless,
 )
@@ -77,6 +80,7 @@ class ProcessedDataFeature:
 class ProcessedData:
     features: list[ProcessedDataFeature]
     identifier: str | None = None
+    custom_info: dict[str, Any] | None = None
 
 
 @dataclass
@@ -102,6 +106,8 @@ class Measurement:
     original_sample_concentration_unit: str | None = None
     baseline_absorbance: float | None = None
     electronic_absorbance_reference_wavelength_setting: float | None = None
+    cursor_position: float | None = None
+    cursor_absorbance: float | None = None
 
     # Measurements
     absorbance: JsonFloat | None = None
@@ -269,6 +275,14 @@ def _get_ultraviolet_absorbance_measurement_document(
                 TQuantityValueMilliAbsorbanceUnit,
                 measurement.baseline_absorbance,
             ),
+            "340 raw": quantity_or_none(
+                TQuantityValueMilliAbsorbanceUnit,
+                (measurement.custom_info or {}).get("340 raw"),
+            ),
+            "Cursor Abs.": quantity_or_none(
+                TQuantityValueMilliAbsorbanceUnit,
+                measurement.cursor_absorbance,
+            ),
         }
         return add_custom_information_document(doc, custom_info_doc)
 
@@ -341,6 +355,9 @@ def _get_device_control_custom_document(
             "dilution factor": quantity_or_none(
                 TQuantityValueUnitless, measurement.dilution_factor_setting
             ),
+            "Cursor Pos.": quantity_or_none(
+                TQuantityValueNanometer, measurement.cursor_position
+            ),
         }
         return (measurement.device_control_custom_info or {}) | custom_info
 
@@ -363,14 +380,19 @@ def _get_sample_document(self, measurement: Measurement) -> SampleDocument:
                 measurement.sample_custom_info["last read standards"]
             )
 
+        sample_custom_info = measurement.sample_custom_info or {}
+        sample_custom_info["Mol. Wt. kda"] = quantity_or_none(
+            TQuantityValueKiloDalton, sample_custom_info.get("Mol. Wt. kda")
+        )
+
         return add_custom_information_document(
             SampleDocument(
                 sample_identifier=measurement.sample_identifier,
                 batch_identifier=measurement.batch_identifier,
                 location_identifier=measurement.location_identifier,
                 well_plate_identifier=measurement.well_plate_identifier,
             ),
-            (measurement.sample_custom_info or {}) | custom_info_doc,
+            sample_custom_info | custom_info_doc,
         )
 
     def _get_processed_data_aggregate_document(
@@ -379,14 +401,38 @@ def _get_processed_data_aggregate_document(
         if not data:
             return None
 
+        # Build custom info for processed data with proper TQuantityValue wrappers
+        processed_data_custom_info = {}
+        if data.custom_info:
+            processed_data_custom_info = {
+                "E1%": quantity_or_none(
+                    TQuantityValueUnitless, data.custom_info.get("E1%")
+                ),
+                "ext. coeff x10e3": quantity_or_none(
+                    TQuantityValuePerMolarPerCentimeter,
+                    data.custom_info.get("ext. coeff x10e3"),
+                ),
+                "ext.c.": quantity_or_none(
+                    TQuantityValuePerMolarPerCentimeter,
+                    data.custom_info.get("ext.c. (l/(mol*cm))"),
+                ),
+                "conc. factor": quantity_or_none(
+                    TQuantityValueNanogramPerMicroliter,
+                    data.custom_info.get("conc. factor (ng/ul)"),
+                ),
+            }
+
         return ProcessedDataAggregateDocument(
             processed_data_document=[
-                ProcessedDataDocumentItem(
-                    # TODO(nstender): figure out how to limit possible classes from get_quantity_class for typing.
-                    mass_concentration=quantity_or_none(
-                        get_quantity_class(feature.unit), feature.result  # type: ignore[arg-type]
+                add_custom_information_document(
+                    ProcessedDataDocumentItem(
+                        # TODO(nstender): figure out how to limit possible classes from get_quantity_class for typing.
+                        mass_concentration=quantity_or_none(
+                            get_quantity_class(feature.unit), feature.result  # type: ignore[arg-type]
+                        ),
+                        processed_data_identifier=data.identifier,
                     ),
-                    processed_data_identifier=data.identifier,
+                    processed_data_custom_info,
                 )
                 for feature in data.features
             ]

src/allotropy/allotrope/schemas/shared/definitions/custom.json

@@ -959,6 +959,26 @@
       }
     ]
   },
+  "tQuantityValuePerMolarPerCentimeter": {
+    "allOf": [
+      {
+        "$ref": "#/$defs/tQuantityValue"
+      },
+      {
+        "$ref": "#/$defs/PerMolarPerCentimeter"
+      }
+    ]
+  },
+  "tNullableQuantityValuePerMolarPerCentimeter": {
+    "allOf": [
+      {
+        "$ref": "#/$defs/tNullableQuantityValue"
+      },
+      {
+        "$ref": "#/$defs/PerMolarPerCentimeter"
+      }
+    ]
+  },
   "tQuantityValuePerMolarPerSecond": {
     "allOf": [
       {

src/allotropy/allotrope/schemas/shared/definitions/units.json

@@ -575,6 +575,18 @@
       "unit"
     ]
   },
+  "PerMolarPerCentimeter": {
+    "properties": {
+      "unit": {
+        "type": "string",
+        "const": "M-1cm-1",
+        "$asm.unit-iri": "http://purl.allotrope.org/ontology/qudt-ext/unit#PerMolarPerCentimeter"
+      }
+    },
+    "required": [
+      "unit"
+    ]
+  },
   "PerMolarPerSecond": {
     "properties": {
       "unit": {

src/allotropy/parsers/thermo_fisher_nanodrop_8000/nanodrop_8000_structure.py

@@ -166,6 +166,14 @@ def create(data: SeriesData) -> SpectroscopyRow:
         )
         unit = data.get(str, "units")
 
+        # Read fields for Data Processing Document custom info
+        processed_data_custom_info = {
+            "E1%": data.get(float, "e 1%"),
+            "ext. coeff x10e3": data.get(float, "ext. coeff x10e3"),
+            "ext.c. (l/(mol*cm))": data.get(float, "ext.c. (l/(mol*cm))"),
+            "conc. factor (ng/ul)": data.get(float, "conc. factor (ng/ul)"),
+        }
+
         measurements: list[Measurement] = []
         for wavelength, absorbance in absorbances.items():
             if absorbance is None:
@@ -190,11 +198,30 @@ def create(data: SeriesData) -> SpectroscopyRow:
                                 unit=unit,
                             )
                         ],
+                        custom_info=processed_data_custom_info or None,
                     )
                     if mass_concentration_capture_wavelength == wavelength
                     and mass_concentration
                     and unit
                     else None,
+                    cursor_position=data.get(float, "cursor pos."),
+                    cursor_absorbance=data.get(float, "cursor abs."),
+                    sample_custom_info={
+                        "Mol. Wt. kda": data.get(float, "mol. wt. kda"),
+                    },
+                    custom_info=data.get_unread(
+                        skip={
+                            "date",
+                            "time",
+                            "user id",
+                            "formula value 1",
+                            "formula name 1",
+                            "formula 1",
+                            "260/230",
+                            "260/280",
+                            "m.w. (g/mol)",
+                        }
+                    ),
                 )
             )