diff --git a/graphique/core.py b/graphique/core.py
index e304c6a..99817cb 100644
--- a/graphique/core.py
+++ b/graphique/core.py
@@ -554,11 +554,10 @@ def get_fragments(self) -> Iterator[ds.Fragment]:
 
     def fragment_keys(self) -> list:
         """Filtered partitioned datasets may not have fragments."""
-        try:
+        with contextlib.suppress(AttributeError, ValueError):
             Table.get_fragments(self)
-        except (AttributeError, ValueError):
-            return []
-        return self.partitioning.schema.names
+            return self.partitioning.schema.names
+        return []
 
     def rank_keys(self, k: int, *names: str, dense: bool = True) -> tuple:
         """Return expression and unmatched fields for partitioned dataset which filters by rank.
diff --git a/graphique/interface.py b/graphique/interface.py
index 3dc192b..a0c7955 100644
--- a/graphique/interface.py
+++ b/graphique/interface.py
@@ -25,7 +25,7 @@
 from .models import Column, doc_field, selections
 from .scalars import Long
 
-Root = Union[ds.Dataset, ds.Scanner, pa.Table]
+Source = Union[ds.Dataset, ds.Scanner, pa.Table]
 
 
 def references(field) -> Iterator:
@@ -61,8 +61,8 @@ class Schema:
 
 @strawberry.interface(description="an arrow dataset, scanner, or table")
 class Dataset:
-    def __init__(self, table: Root):
-        self.table = table
+    def __init__(self, source: Source):
+        self.source = source
 
     def references(self, info: Info, level: int = 0) -> set:
         """Return set of every possible future column reference."""
@@ -74,7 +74,7 @@ def references(self, info: Info, level: int = 0) -> set:
     def scanner(self, info: Info, **options) -> ds.Scanner:
         """Return scanner with only the columns necessary to proceed."""
         options.setdefault('columns', list(self.references(info)))
-        dataset = ds.dataset(self.table) if isinstance(self.table, pa.Table) else self.table
+        dataset = ds.dataset(self.source) if isinstance(self.source, pa.Table) else self.source
         if isinstance(dataset, ds.Dataset):
             return dataset.scanner(**options)
         options['schema'] = dataset.projected_schema
@@ -82,8 +82,8 @@ def scanner(self, info: Info, **options) -> ds.Scanner:
 
     def select(self, info: Info, length: Optional[int] = None) -> pa.Table:
         """Return table with only the rows and columns necessary to proceed."""
-        if isinstance(self.table, pa.Table):
-            return self.table.select(self.references(info))
+        if isinstance(self.source, pa.Table):
+            return self.source.select(self.references(info))
         scanner = self.scanner(info)
         if length is None:
             return self.add_metric(info, scanner.to_table(), mode='read')
@@ -118,16 +118,16 @@ def filter(self, info: Info, **queries: Filter) -> Self:
         See `scan(filter: ...)` for more advanced queries. Additional feature: sorted tables
         support binary search
         """
-        table = self.table
+        source = self.source
         prev = info.path.prev
-        search = isinstance(table, pa.Table) and (prev is None or prev.typename == 'Query')
+        search = isinstance(source, pa.Table) and (prev is None or prev.typename == 'Query')
         for name in self.schema().index if search else []:
-            assert not table[name].null_count, f"search requires non-null column: {name}"
+            assert not source[name].null_count, f"search requires non-null column: {name}"
             query = dict(queries.pop(name))
             if 'eq' in query:
-                table = T.is_in(table, name, *query['eq'])
+                source = T.is_in(source, name, *query['eq'])
             if 'ne' in query:
-                table = T.not_equal(table, name, query['ne'])
+                source = T.not_equal(source, name, query['ne'])
             lower, upper = query.get('gt'), query.get('lt')
             includes = {'include_lower': False, 'include_upper': False}
             if 'ge' in query and (lower is None or query['ge'] > lower):
@@ -135,24 +135,24 @@ def filter(self, info: Info, **queries: Filter) -> Self:
             if 'le' in query and (upper is None or query['le'] > upper):
                 upper, includes['include_upper'] = query['le'], True
             if {lower, upper} != {None}:
-                table = T.range(table, name, lower, upper, **includes)
+                source = T.range(source, name, lower, upper, **includes)
             if len(query.pop('eq', [])) != 1 or query:
                 break
-        self = type(self)(table)
+        self = type(self)(source)
         expr = Expression.from_query(**queries)
         return self if expr.to_arrow() is None else self.scan(info, filter=expr)
 
     @doc_field
     def type(self) -> str:
         """[arrow type](https://arrow.apache.org/docs/python/api/dataset.html#classes)"""
-        return type(self.table).__name__
+        return type(self.source).__name__
 
     @doc_field
     def schema(self) -> Schema:
         """dataset schema"""
-        table = self.table
-        schema = table.projected_schema if isinstance(table, ds.Scanner) else table.schema
-        partitioning = getattr(table, 'partitioning', None)
+        source = self.source
+        schema = source.projected_schema if isinstance(source, ds.Scanner) else source.schema
+        partitioning = getattr(source, 'partitioning', None)
         index = (schema.pandas_metadata or {}).get('index_columns', [])
         return Schema(
             names=schema.names,
@@ -184,7 +184,7 @@ def add_metric(info: Info, table: pa.Table, **data):
     @doc_field
     def length(self) -> Long:
         """number of rows"""
-        return len(self.table) if hasattr(self.table, '__len__') else self.table.count_rows()
+        return len(self.source) if hasattr(self.source, '__len__') else self.source.count_rows()
 
     @doc_field
     def any(self, info: Info, length: Long = 1) -> bool:
@@ -198,7 +198,7 @@ def any(self, info: Info, length: Long = 1) -> bool:
     @doc_field
     def size(self) -> Optional[Long]:
         """buffer size in bytes; null if table is not loaded"""
-        return getattr(self.table, 'nbytes', None)
+        return getattr(self.source, 'nbytes', None)
 
     @doc_field(
         name="column name(s); multiple names access nested struct fields",
@@ -248,24 +248,24 @@ def group(
         See `column` for accessing any column which has changed type. See `tables` to split on any
         aggregated list columns.
         """
-        table, aggs = self.table, dict(aggregate)
+        source, aggs = self.source, dict(aggregate)
         refs = {agg.name for values in aggs.values() for agg in values}
-        fragments = set(T.fragment_keys(self.table))
-        if isinstance(table, ds.Scanner):
-            table = self.select(info)
+        fragments = set(T.fragment_keys(self.source))
+        if isinstance(source, ds.Scanner):
+            source = self.select(info)
         if fragments and set(by) <= fragments:
             if set(by) == fragments:
                 return type(self)(self.fragments(info, counts, aggregate))
             if fragments.isdisjoint(refs) and set(aggs) <= Field.associatives:
-                table = self.fragments(info, counts, aggregate)
+                source = self.fragments(info, counts, aggregate)
                 aggs.setdefault('sum', []).extend(Field(agg.alias) for agg in aggs.pop('count', []))
                 if counts:
                     aggs['sum'].append(Field(counts))
                     counts = ''
                 for agg in itertools.chain(*aggs.values()):
                     agg.name = agg.alias
-        loaded = isinstance(table, pa.Table)
-        table = T.group(table, *by, counts=counts, ordered=ordered, **aggs)
+        loaded = isinstance(source, pa.Table)
+        table = T.group(source, *by, counts=counts, ordered=ordered, **aggs)
         return type(self)(table if loaded else self.add_metric(info, table, mode='group'))
 
     def fragments(self, info: Info, counts: str = '', aggregate: HashAggregates = {}) -> pa.Table:  # type: ignore
@@ -273,13 +273,13 @@ def fragments(self, info: Info, counts: str = '', aggregate: HashAggregates = {}
 
         Requires a partitioned dataset. Faster and less memory intensive than `group`.
         """
-        schema = self.table.partitioning.schema  # requires a Dataset
+        schema = self.source.partitioning.schema  # requires a Dataset
         aggs = dict(aggregate)
         names = self.references(info, level=1)
         names.update(agg.name for value in aggs.values() for agg in value)
         projection = {name: pc.field(name) for name in names - set(schema.names)}
         columns = collections.defaultdict(list)
-        for fragment in T.get_fragments(self.table):
+        for fragment in T.get_fragments(self.source):
             row = ds.get_partition_keys(fragment.partition_expression)
             if projection:
                 table = fragment.to_table(columns=projection)
@@ -338,10 +338,10 @@ def sort(
         Optimized for length == 1; matches min or max values.
         """
         kwargs = dict(length=length, null_placement=null_placement)
-        if isinstance(self.table, pa.Table) or length is None:
+        if isinstance(self.source, pa.Table) or length is None:
             table = self.select(info)
         else:
-            expr, by = T.rank_keys(self.table, length, *by, dense=False)
+            expr, by = T.rank_keys(self.source, length, *by, dense=False)
             scanner = self.scanner(info, filter=expr)
             if not by:
                 return type(self)(self.add_metric(info, scanner.head(length), mode='head'))
@@ -355,13 +355,13 @@ def sort(
     )
     def rank(self, info: Info, by: list[str], max: int = 1) -> Self:
         """Return table selected by maximum dense rank."""
-        expr, by = T.rank_keys(self.table, max, *by)
-        table = self.table if expr is None else self.table.filter(expr)
+        expr, by = T.rank_keys(self.source, max, *by)
+        source = self.source if expr is None else self.source.filter(expr)
         if not by:
-            return type(self)(table)
-        if not isinstance(table, ds.Dataset):
-            table = self.select(info)
-        return type(self)(T.rank(table, max, *by))
+            return type(self)(source)
+        if not isinstance(source, ds.Dataset):
+            source = self.select(info)
+        return type(self)(T.rank(source, max, *by))
 
     @staticmethod
     def apply_list(table: Batch, list_: ListFunction) -> Batch:
@@ -500,10 +500,10 @@ def oneshot(cls, info: Info, scanner: ds.Scanner) -> Union[ds.Scanner, pa.Table]
     def scan(self, info: Info, filter: Expression = {}, columns: list[Projection] = []) -> Self:  # type: ignore
         """Select rows and project columns without memory usage."""
         expr = filter.to_arrow()
-        if expr is not None and not columns and isinstance(self.table, ds.Dataset):
-            return type(self)(self.table.filter(expr))
+        if expr is not None and not columns and isinstance(self.source, ds.Dataset):
+            return type(self)(self.source.filter(expr))
         scanner = self.scanner(info, filter=expr, columns=self.project(info, columns))
-        if isinstance(self.table, ds.Scanner):
+        if isinstance(self.source, ds.Scanner):
             scanner = self.oneshot(info, scanner)
         return type(self)(scanner)
 
@@ -529,7 +529,7 @@ def join(
     ) -> Self:
         """Provisional: [join](https://arrow.apache.org/docs/python/generated/pyarrow.dataset.Dataset.html#pyarrow.dataset.Dataset.join) this table with another table on the root Query type."""
         left, right = (
-            root.table if isinstance(root.table, ds.Dataset) else root.select(info)
+            root.source if isinstance(root.source, ds.Dataset) else root.select(info)
             for root in (self, getattr(info.root_value, right))
         )
         table = left.join(
@@ -554,7 +554,7 @@ def take(self, info: Info, indices: list[Long]) -> Self:
     @doc_field
     def drop_null(self, info: Info) -> Self:
         """Remove missing values from referenced columns in the table."""
-        if isinstance(self.table, pa.Table):
+        if isinstance(self.source, pa.Table):
             return type(self)(pc.drop_null(self.select(info)))
         scanner = self.scanner(info)
         batches = map(pc.drop_null, scanner.to_batches())
diff --git a/graphique/middleware.py b/graphique/middleware.py
index b9523a6..970b9f3 100644
--- a/graphique/middleware.py
+++ b/graphique/middleware.py
@@ -13,7 +13,7 @@
 from strawberry.extensions import tracing
 from strawberry.utils.str_converters import to_camel_case
 from .inputs import Filter
-from .interface import Dataset, Root
+from .interface import Dataset, Source
 from .models import Column, doc_field
 from .scalars import Long, py_type, scalar_map
 
@@ -48,7 +48,7 @@ class GraphQL(strawberry.asgi.GraphQL):
 
     options = dict(types=Column.registry.values(), scalar_overrides=scalar_map)
 
-    def __init__(self, root: Root, debug: bool = False, **kwargs):
+    def __init__(self, root: Source, debug: bool = False, **kwargs):
         options: dict = dict(self.options, extensions=(MetricsExtension,) * bool(debug))
         if type(root).__name__ == 'Query':
             self.root_value = root
@@ -63,7 +63,7 @@ async def get_root_value(self, request):
         return self.root_value
 
     @classmethod
-    def federated(cls, roots: Mapping[str, Root], keys: Mapping[str, Iterable] = {}, **kwargs):
+    def federated(cls, roots: Mapping[str, Source], keys: Mapping[str, Iterable] = {}, **kwargs):
         """Construct GraphQL app with multiple federated datasets.
 
         Args:
@@ -77,7 +77,7 @@ def federated(cls, roots: Mapping[str, Root], keys: Mapping[str, Iterable] = {},
         return cls(strawberry.type(Query)(**root_values), **kwargs)
 
 
-def implemented(root: Root, name: str = '', keys: Iterable = ()):
+def implemented(root: Source, name: str = '', keys: Iterable = ()):
     """Return type which extends the Dataset interface with knowledge of the schema."""
     schema = root.projected_schema if isinstance(root, ds.Scanner) else root.schema
     types = {field.name: py_type(field.type) for field in schema}
diff --git a/graphique/service.py b/graphique/service.py
index 3ac9f0b..94f4141 100644
--- a/graphique/service.py
+++ b/graphique/service.py
@@ -9,6 +9,7 @@
 
 import json
 from pathlib import Path
+import pyarrow as pa
 import pyarrow.dataset as ds
 from starlette.config import Config
 from graphique.inputs import Expression
@@ -25,6 +26,9 @@
 
 if isinstance(COLUMNS, dict):
     COLUMNS = {alias: ds.field(name) for alias, name in COLUMNS.items()}
+elif COLUMNS:
+    root = root.replace_schema(pa.schema(map(root.schema.field, COLUMNS), root.schema.metadata))
+    COLUMNS = None
 if FILTERS is not None:
     root = root.to_table(columns=COLUMNS, filter=Expression.from_query(**FILTERS).to_arrow())
 elif COLUMNS: