jfaustin's picture
add dockerfile and folding studio cli
44459bb
raw
history blame
6.26 kB
"""Utilitaries methods for the commands module."""
import hashlib
from contextlib import contextmanager
from rich.progress import Progress, SpinnerColumn, TextColumn
from folding_studio.api_call.upload_custom_files import (
CustomFileType,
upload_custom_files,
)
from folding_studio.console import console
@contextmanager
def success_fail_catch_spinner(message: str, spinner_name: str = "dots"):
"""Wrapper around a rich progress spinner that adapt its state icon.
Args:
message (str): message to show supporting rich format.
spinner_name (str, optional): rich SpinnerColumn spinner_name attribute. Defaults to "dots".
Examples:
```
with success_fail_catch_spinner("Running Task"):
...
>>> Running Task β Ό # spins as long as the context manager is running
# if no error raised then transforms into
>>> Running Task βœ…
# otherwise transforms into
>>> Running Task ❌
An error occurred: <ERROR>
...
```
"""
err = None
with Progress(
TextColumn("{task.description}"),
SpinnerColumn(spinner_name, finished_text=""),
console=console,
) as progress:
task_id = progress.add_task(message, total=1)
# response = client.send_request(query)
try:
yield
progress.update(
task_id, completed=1, description=f"{message} :white_check_mark:"
)
except Exception as e:
progress.update(task_id, completed=1, description=f"{message} :x:")
err = e
# for coherent message order the print has to be made outside the Progress context manager
if err is not None:
console.print(f"An error occurred: {err}")
raise err
@contextmanager
def success_fail_catch_print(*args, **kwargs):
"""Wrapper around rich `print` that adapts its state icon.
Examples:
```
with success_fail_catch_print("Running Task..."):
...
>>> Running Task...
# if no error raised then transforms into
>>> Running Task... βœ…
# otherwise transforms into
>>> Running Task... ❌
An error occurred: <ERROR>
...
```
"""
console.print(*args, **kwargs, end=" ")
try:
yield
console.print(":white_check_mark:")
except Exception as e:
console.print(":x:")
console.print(f"An error occurred: {e}")
raise e
def a3m_to_aligned_pqt(directory: str) -> str:
"""
Finds .a3m files in a directory and merges them into a single aligned Parquet file.
Args:
directory (str): Path to the directory containing .a3m files.
Returns:
str: The path to the saved Parquet file.
Raises:
ValueError: If the directory is invalid, if no records are found in a file,
or if query sequences differ among files.
"""
dir_path = Path(directory)
if not dir_path.is_dir():
raise ValueError(f"{directory} is not a valid directory.")
mapped_files = {}
for file in dir_path.glob("*.a3m"):
dbname = file.stem.replace("_hits", "").replace("hits_", "")
source = dbname.lower() if dbname else "uniref90"
mapped_files[file] = source
def parse_a3m(file_path: Path, source: str) -> pd.DataFrame:
"""
Parses a simple FASTA file.
The first record is flagged with source "query"; subsequent records use the provided source.
Uses the header both as a comment and (if desired) as a pairing key.
"""
with open(file_path, "r") as f:
lines = f.read().splitlines()
records = []
header = None
seq_lines = []
for line in lines:
if line.startswith(">"):
if header is not None:
seq = "".join(seq_lines).strip()
record_source = "query" if not records else source
records.append(
{
"sequence": seq,
"source_database": record_source,
"pairing_key": header,
"comment": header,
}
)
header = line[1:].strip()
seq_lines = []
else:
seq_lines.append(line.strip())
if header is not None:
seq = "".join(seq_lines).strip()
record_source = "query" if not records else source
records.append(
{
"sequence": seq,
"source_database": record_source,
"pairing_key": header,
"comment": header,
}
)
if not records:
raise ValueError(f"No records found in {file_path}")
return pd.DataFrame.from_records(records)
dfs = {}
for file, source in mapped_files.items():
dfs[file] = parse_a3m(file, source)
query_set = {df.iloc[0]["sequence"] for df in dfs.values()}
if len(query_set) != 1:
raise ValueError("Query sequences differ among files.")
merged_df = None
for df in dfs.values():
if merged_df is None:
merged_df = df.iloc[0:1].copy()
merged_df = pd.concat([merged_df, df.iloc[1:]], ignore_index=True)
query_seq = merged_df.iloc[0]["sequence"]
def hash_sequence(seq: str) -> str:
return hashlib.sha256(seq.upper().encode()).hexdigest()
output_filename = f"{hash_sequence(query_seq)}.aligned.pqt"
dir_path.mkdir(exist_ok=True, parents=True)
out_path = dir_path / output_filename
merged_df.to_parquet(out_path, index=False)
return str(out_path)
def process_uploaded_msas(msa_files, headers):
"""
Uploads the given MSA files and returns a dictionary mapping file names to their uploaded values.
"""
uploaded = upload_custom_files(
headers=headers, paths=msa_files, file_type=CustomFileType.MSA
)
msa_paths = {}
for f in msa_files:
msa_paths[f.name] = uploaded.get(str(f)) or uploaded.get(f.name)
return msa_paths