diff --git a/src/io/tdf/mod.rs b/src/io/tdf/mod.rs
index df85a8d..44c8d3d 100644
--- a/src/io/tdf/mod.rs
+++ b/src/io/tdf/mod.rs
@@ -10,4 +10,5 @@ mod arrays;
 mod sql;
 mod reader;
 
-pub use reader::{TDFFrameReader, TDFFrameReaderType, TDFSpectrumReader, TDFSpectrumReaderType, is_tdf};
\ No newline at end of file
+pub use reader::{TDFFrameReader, TDFFrameReaderType, TDFSpectrumReader, TDFSpectrumReaderType, is_tdf};
+pub use sql::{ChromatographyData, SQLTrace};
\ No newline at end of file
diff --git a/src/io/tdf/reader.rs b/src/io/tdf/reader.rs
index be4f756..966965c 100644
--- a/src/io/tdf/reader.rs
+++ b/src/io/tdf/reader.rs
@@ -46,13 +46,13 @@ use timsrust::{
     Metadata, TimsRustError,
 };
 
-use super::arrays::consolidate_peaks;
 pub use super::arrays::FrameToArraysMapper;
 use super::constants::{InstrumentSource, MsMsType};
 use super::sql::{
     FromSQL, PasefPrecursor, RawTDFSQLReader, SQLDIAFrameMsMsWindow, SQLFrame, SQLPasefFrameMsMs,
     SQLPrecursor, TDFMSnFacet,
 };
+use super::{arrays::consolidate_peaks, sql::ChromatographyData};
 
 const PEAK_MERGE_TOLERANCE: Tolerance = Tolerance::Da(0.01);
 
@@ -174,10 +174,8 @@ pub struct TDFFrameReaderType<
     _d: PhantomData<D>,
 }
 
-
 type DIAFrameWindowMap = HashMap<u32, Vec<Arc<SQLDIAFrameMsMsWindow>>, BuildIdentityHasher<u32>>;
 
-
 impl<C: FeatureLike<MZ, IonMobility>, D: FeatureLike<Mass, IonMobility> + KnownCharge>
     TDFFrameReaderType<C, D>
 {
@@ -316,10 +314,7 @@ impl<C: FeatureLike<MZ, IonMobility>, D: FeatureLike<Mass, IonMobility> + KnownC
         Ok((entry_count_guess, frame_types))
     }
 
-    fn build_window_index(
-        &self,
-    ) -> Result<DIAFrameWindowMap, Error>
-    {
+    fn build_window_index(&self) -> Result<DIAFrameWindowMap, Error> {
         let conn = self.tdf_reader.connection();
         let dia_windows = SQLDIAFrameMsMsWindow::read_from(&conn, [])?;
         let mut window_groups: HashMap<u32, Vec<Arc<SQLDIAFrameMsMsWindow>>, _> =
@@ -450,8 +445,7 @@ impl<C: FeatureLike<MZ, IonMobility>, D: FeatureLike<Mass, IonMobility> + KnownC
                         .extend_from_slice(&entry.frame.summed_intensities.to_le_bytes());
                 }
                 ChromatogramType::BasePeakChromatogram => {
-                    intensity_array
-                        .extend_from_slice(&entry.frame.max_intensity.to_le_bytes());
+                    intensity_array.extend_from_slice(&entry.frame.max_intensity.to_le_bytes());
                 }
                 _ => {
                     unimplemented!()
@@ -597,6 +591,21 @@ impl<C: FeatureLike<MZ, IonMobility>, D: FeatureLike<Mass, IonMobility> + KnownC
             Ok(None)
         }
     }
+
+    pub fn get_trace_reader(&self) -> Result<ChromatographyData, TimsRustError> {
+        let path = self
+            .metadata
+            .path
+            .parent()
+            .expect(".tdf file did not have an enclosing directory")
+            .join(super::sql::ChromatographyData::FILE_NAME);
+        let handle = super::sql::ChromatographyData::new(&path).map_err(|e| {
+            TimsRustError::MetadataReaderError(timsrust::readers::MetadataReaderError::SqlError(
+                e.into(),
+            ))
+        })?;
+        Ok(handle)
+    }
 }
 
 // Metadata construction routine
@@ -1249,6 +1258,10 @@ impl<
     pub fn peak_merging_tolerance_mut(&mut self) -> &mut Tolerance {
         &mut self.peak_merging_tolerance
     }
+
+    pub fn get_trace_reader(&self) -> Result<ChromatographyData, TimsRustError> {
+        self.frame_reader.get_trace_reader()
+    }
 }
 
 pub type TDFSpectrumReader = TDFSpectrumReaderType<
@@ -1277,7 +1290,6 @@ fn index_to_precursor(
 
         let im = metadata.im_converter.convert(prec.scan_average);
 
-
         let p = inverse_reduce_ion_mobility_param(im);
         ion.add_param(p);
         let mut act = Activation::default();
diff --git a/src/io/tdf/sql.rs b/src/io/tdf/sql.rs
index 5f7a9af..2460bdd 100644
--- a/src/io/tdf/sql.rs
+++ b/src/io/tdf/sql.rs
@@ -494,3 +494,202 @@ impl TDFMSnFacet {
         }
     }
 }
+
+#[derive(Debug, Default, Clone, PartialEq)]
+pub struct SQLTrace {
+    pub id: usize,
+    pub description: String,
+    pub instrument: String,
+    pub instrument_id: String,
+    pub trace_type: usize,
+    pub unit: usize,
+    pub time_offset: f64,
+    pub times: Option<Vec<f64>>,
+    pub intensities: Option<Vec<f32>>,
+}
+
+impl SQLTrace {
+    #[allow(clippy::too_many_arguments)]
+    pub fn new(
+        id: usize,
+        description: String,
+        instrument: String,
+        instrument_id: String,
+        trace_type: usize,
+        unit: usize,
+        time_offset: f64,
+        times: Option<Vec<f64>>,
+        intensities: Option<Vec<f32>>,
+    ) -> Self {
+        Self {
+            id,
+            description,
+            instrument,
+            instrument_id,
+            trace_type,
+            unit,
+            time_offset,
+            times,
+            intensities,
+        }
+    }
+
+    pub fn load_trace_data(&mut self, connection: &Connection) -> Result<&Self, Error> {
+        let mut q =
+            connection.prepare("SELECT Times, Intensities FROM TraceChunks WHERE Id = ?")?;
+        let (mut times, mut intensities) = q
+            .query([self.id])?
+            .mapped(|row| {
+                let times = row.get::<usize, Vec<u8>>(0)?;
+                let mut time_acc = Vec::new();
+                let mut buf = [0u8; 8];
+                for chunk in times.chunks_exact(8) {
+                    buf.copy_from_slice(chunk);
+                    time_acc.push(f64::from_le_bytes(buf));
+                }
+
+                let ints = row.get::<usize, Vec<u8>>(1)?;
+                let mut intensity_acc = Vec::new();
+                let mut buf = [0u8; 4];
+                for chunk in ints.chunks_exact(4) {
+                    buf.copy_from_slice(chunk);
+                    intensity_acc.push(f32::from_le_bytes(buf));
+                }
+
+                Ok((time_acc, intensity_acc))
+            })
+            .flatten()
+            .reduce(|(mut ta, mut ia), (t, i)| {
+                ta.extend_from_slice(&t);
+                ia.extend_from_slice(&i);
+                (ta, ia)
+            })
+            .unwrap_or_default();
+
+        if !times.is_sorted() {
+            let mut indices: Vec<_> = (0..times.len()).collect();
+            indices.sort_by(|a, b| times[*a].total_cmp(&times[*b]));
+            const TOMBSTONE: usize = usize::MAX;
+
+            for idx in 0..times.len() {
+                if indices[idx] != TOMBSTONE {
+                    let mut current_idx = idx;
+                    loop {
+                        let next_idx = indices[current_idx];
+                        indices[current_idx] = TOMBSTONE;
+                        if indices[next_idx] == TOMBSTONE {
+                            break;
+                        }
+                        times.swap(current_idx, next_idx);
+                        intensities.swap(current_idx, next_idx);
+                        current_idx = next_idx;
+                    }
+                }
+            }
+        }
+
+        self.times = Some(times);
+        self.intensities = Some(intensities);
+        Ok(self)
+    }
+
+    pub fn len(&self) -> usize {
+        self.times.as_ref().map(|s| s.len()).unwrap_or_default()
+    }
+
+    pub fn is_empty(&self) -> bool {
+        self.times
+            .as_ref()
+            .map(|s| s.is_empty())
+            .unwrap_or_default()
+    }
+
+    pub fn iter(&self) -> impl Iterator<Item = (f64, f32)> + '_ {
+        self.times
+            .as_ref()
+            .into_iter()
+            .flatten()
+            .copied()
+            .zip(self.intensities.as_ref().into_iter().flatten().copied())
+    }
+}
+
+impl FromSQL for SQLTrace {
+    fn from_row(row: &Row<'_>) -> Result<Self, Error> {
+        let this = Self::new(
+            row.get(0)?,
+            row.get(1)?,
+            row.get(2)?,
+            row.get(3)?,
+            row.get(4)?,
+            row.get(5)?,
+            row.get(6)?,
+            None,
+            None,
+        );
+        Ok(this)
+    }
+
+    fn get_sql() -> String {
+        "SELECT Id, Description, Instrument, InstrumentId, Type, Unit, TimeOffset FROM TraceSources"
+            .into()
+    }
+}
+
+// chromatography-data.sqlite
+pub struct ChromatographyData {
+    pub connection: ReentrantMutex<Connection>,
+}
+
+impl ChromatographyData {
+    pub const FILE_NAME: &str = "chromatography-data.sqlite";
+
+    pub fn new(data_path: &Path) -> Result<Self, Error> {
+        let connection = ReentrantMutex::new(Connection::open(data_path)?);
+        let this = Self { connection };
+        this.valid_schema()?;
+        Ok(this)
+    }
+
+    fn valid_schema(&self) -> Result<bool, Error> {
+        Ok(self.has_table("TraceSources")? && self.has_table("TraceChunks")?)
+    }
+
+    pub fn connection(&self) -> ReentrantMutexGuard<'_, Connection> {
+        self.connection.try_lock().unwrap()
+    }
+
+    fn has_table(&self, table: &str) -> Result<bool, Error> {
+        self.connection().query_row(
+            "SELECT name FROM sqlite_master WHERE type='table' AND name=?",
+            [table],
+            |_| Ok(true),
+        )
+    }
+
+    pub fn traces(&mut self, include_data: bool) -> Result<Vec<SQLTrace>, Error> {
+        let mut traces = SQLTrace::read_from(&self.connection(), [])?;
+        if include_data {
+            for t in traces.iter_mut() {
+                t.load_trace_data(&self.connection())?;
+            }
+        }
+        Ok(traces)
+    }
+
+    pub fn get_by_name(
+        &mut self,
+        name: &str,
+        include_data: bool,
+    ) -> Result<Option<SQLTrace>, Error> {
+        let mut trace = SQLTrace::read_from_where(&self.connection(), [name], "Description = ?")?
+            .into_iter()
+            .next();
+        if include_data {
+            if let Some(t) = trace.as_mut() {
+                t.load_trace_data(&self.connection())?;
+            }
+        }
+        Ok(trace)
+    }
+}