refactor(eval): reuse shared evo2_sae.fasta.read_fasta in evo2_buffer

Drop evo2_buffer's local read_fasta — it already imports evo2_sae, so reuse the shared
reader (also handles .gz). Kingdom check uses the id prefix, which the shared reader preserves.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: Polina Binder <pbinder@nvidia.com>

Author: polinabinder1

bionemo-recipes/interpretability/sparse_autoencoders/recipes/evo2/scripts/evo2_buffer.py

@@ -28,27 +28,13 @@
 import labelers as L
 import numpy as np
 import torch
+from evo2_sae.fasta import read_fasta  # shared reader (also handles .gz); kingdom check uses the id prefix
 from sae.eval.probing import ActivationBuffer
 
 
 KINGDOM_TAGS = {"prok": "|d__Bacteria|", "euk": "|d__Eukaryota|"}
 
 
-def read_fasta(path):  # noqa: D103
-    header, chunks = None, []
-    with open(path) as fh:
-        for line in fh:
-            line = line.rstrip("\n")
-            if line.startswith(">"):
-                if header is not None:
-                    yield header, "".join(chunks)
-                header, chunks = line[1:], []
-            else:
-                chunks.append(line)
-    if header is not None:
-        yield header, "".join(chunks)
-
-
 def sample_sequences(fasta, max_tokens, seq_len, kingdoms=("prok", "euk"), seed=0):  # noqa: D103
     from evo2_sae.core import clean_dna