|
| 1 | +from urllib.request import urlopen |
| 2 | +from urllib.error import URLError |
| 3 | +import csv |
| 4 | +import sys |
| 5 | +from ..simple_table import Table |
| 6 | +from ..loader import IMGT_HLA_URL |
| 7 | + |
| 8 | + |
| 9 | +def load_allele_list(imgt_version): |
| 10 | + """ |
| 11 | + The format of the AlleleList file has a 6-line header with a header |
| 12 | + on the 7th line |
| 13 | + ``` |
| 14 | + # file: Allelelist.3290.txt |
| 15 | + # date: 2017-07-10 |
| 16 | + # version: IPD-IMGT/HLA 3.29.0 |
| 17 | + # origin: https://github.com/ANHIG/IMGTHLA/Allelelist.3290.txt |
| 18 | + # repository: https://raw.githubusercontent.com/ANHIG/IMGTHLA/Latest/allelelist/Allelelist.3290.txt |
| 19 | + # author: WHO, Steven G. E. Marsh ([email protected]) |
| 20 | + AlleleID,Allele |
| 21 | + HLA00001,A*01:01:01:01 |
| 22 | + HLA02169,A*01:01:01:02N |
| 23 | + HLA14798,A*01:01:01:03 |
| 24 | + HLA15760,A*01:01:01:04 |
| 25 | + HLA16415,A*01:01:01:05 |
| 26 | + HLA16417,A*01:01:01:06 |
| 27 | + HLA16436,A*01:01:01:07 |
| 28 | + ``` |
| 29 | +
|
| 30 | + Returns a Table object with AlleleID and Allele columns |
| 31 | +
|
| 32 | + :param imgt_version: IMGT database version |
| 33 | + :return: Table object with AlleleID and Allele data |
| 34 | + """ |
| 35 | + |
| 36 | + if imgt_version == "Latest": |
| 37 | + allele_list_url = f"{IMGT_HLA_URL}Latest/Allelelist.txt" |
| 38 | + else: |
| 39 | + if imgt_version == "3130": |
| 40 | + # 3130 was renamed to 3131 for Allelelist file only 🤷🏾 |
| 41 | + imgt_version = "3131" |
| 42 | + allele_list_url = ( |
| 43 | + f"{IMGT_HLA_URL}Latest/allelelist/Allelelist.{imgt_version}.txt" |
| 44 | + ) |
| 45 | + |
| 46 | + try: |
| 47 | + response = urlopen(allele_list_url) |
| 48 | + lines = [line.decode("utf-8").strip() for line in response] |
| 49 | + |
| 50 | + # Skip first 6 header lines |
| 51 | + data_lines = lines[6:] |
| 52 | + |
| 53 | + reader = csv.DictReader(data_lines) |
| 54 | + columns = ["AlleleID", "Allele"] |
| 55 | + |
| 56 | + return Table(reader, columns) |
| 57 | + except URLError as e: |
| 58 | + print(f"Error downloading {allele_list_url}", e, file=sys.stderr) |
| 59 | + sys.exit(1) |
0 commit comments