From f4f81a18c5f6c05558fe539261f42e43c5202125 Mon Sep 17 00:00:00 2001
From: Martin Larralde <martin.larralde@embl.de>
Date: Tue, 8 Oct 2024 20:59:11 +0200
Subject: [PATCH] Fix detection of alphabet from arbitrary queries in `phmmer`
and `hmmscan`
---
pyhmmer/hmmer.py | 31 ++++++++++++++++++++-----------
1 file changed, 20 insertions(+), 11 deletions(-)
diff --git a/pyhmmer/hmmer.py b/pyhmmer/hmmer.py
index 05c511c0..fa1e1fe9 100644
--- a/pyhmmer/hmmer.py
+++ b/pyhmmer/hmmer.py
@@ -995,14 +995,13 @@ def phmmer(
alphabet = alphabet or sequences.alphabet
targets = sequences
else:
- alphabet = alphabet or Alphabet.amino()
- targets = DigitalSequenceBlock(alphabet, sequences)
-
- if builder is None:
- builder = Builder(
- alphabet,
- seed=options.get("seed", 42)
- )
+ sequences = peekable(sequences)
+ try:
+ alphabet = alphabet or sequences.peek().alphabet or Alphabet.amino()
+ targets = DigitalSequenceBlock(alphabet, sequences)
+ except StopIteration:
+ alphabet = alphabet or Alphabet.amino()
+ targets = DigitalSequenceBlock(alphabet)
if "alphabet" not in options:
options["alphabet"] = alphabet
@@ -1170,8 +1169,13 @@ def jackhmmer(
alphabet = alphabet or sequences.alphabet
targets = sequences
else:
- alphabet = alphabet or Alphabet.amino()
- targets = DigitalSequenceBlock(alphabet, sequences)
+ sequences = peekable(sequences)
+ try:
+ alphabet = alphabet or sequences.peek().alphabet or Alphabet.amino()
+ targets = DigitalSequenceBlock(alphabet, sequences)
+ except StopIteration:
+ alphabet = alphabet or Alphabet.amino()
+ targets = DigitalSequenceBlock(alphabet)
if builder is None:
builder = Builder(
@@ -1513,7 +1517,12 @@ def hmmscan(
if not isinstance(queries, collections.abc.Iterable):
queries = (queries,)
if isinstance(profiles, HMMPressedFile):
- alphabet = alphabet or Alphabet.amino() # FIXME: try to detect alphabet?
+ opt = profiles.read()
+ profiles.rewind()
+ if opt is not None:
+ alphabet = alphabet or opt.alphabet
+ else:
+ alphabet = Alphabet.amino()
targets = profiles
elif isinstance(profiles, OptimizedProfileBlock):
alphabet = alphabet or profiles.alphabet
--
GitLab