diff --git a/setup.cfg b/setup.cfg
index a009334..dca5be2 100644
--- a/setup.cfg
+++ b/setup.cfg
@@ -64,6 +64,7 @@ install_requires =
     hotelling
 #    fastsk
     upsetplot
+    geosketch
 
 
 [options.packages.find]
diff --git a/src/polygraph/embedding.py b/src/polygraph/embedding.py
index f34b136..99b779e 100644
--- a/src/polygraph/embedding.py
+++ b/src/polygraph/embedding.py
@@ -414,6 +414,44 @@ def distribution_shift(ad, reference_group, group_col="Group", use_pca=False):
     return ad
 
 
+def geometric_sketch(ad, N, groups=None, group_col="Group", use_pca=True):
+    """
+    Applies geometric sketching (Hie, Brian et al. Cell Systems, Volume 8,
+    Issue 6, 483 - 493.e7) to sample a subset of sequences that represent
+    the diversity in the specified groups.
+
+    Args:
+        ad (anndata.AnnData): Anndata object containing sequence embeddings
+            of shape (n_seqs x n_vars)
+        N (int): Number of sequences to sample from each group
+        groups (list): Names of groups from which to sample sequences. If None,
+            all groups are used.
+        group_col (str): Name of column in .obs containing group ID
+        use_pca (bool): Whether to use PCA distances
+
+    Returns:
+        ad (anndata.AnnData): Modified anndata object containing selections in
+            ad.obs['selected'].
+    """
+    from geosketch import gs
+
+    ad.obs["selected"] = False
+    groups = groups or ad.obs[group_col].unique()
+
+    for group in groups:
+        in_group = ad.obs[group_col] == group
+        group_idx = ad.obs_names[in_group].tolist()
+        if use_pca:
+            group_X = ad.obsm["X_pca"][in_group, :]
+        else:
+            group_X = ad.X[in_group, :]
+
+        sketch_index = gs(group_X, N=N, replace=False)
+        ad.obs.loc[[group_idx[x] for x in sketch_index], "selected"] = True
+
+    return ad
+
+
 def embedding_analysis(
     matrix,
     seqs,