Merge pull request #183 from SPAAM-community/nfcore_mag_samplesheet

Incorporate the changes in the nf-core/mag samplesheet for its generation

Author: alexhbnr

amdirt/core/__init__.py

@@ -233,8 +233,12 @@ def get_filename(path_string: str, orientation: str) -> Tuple[str, str]:
     """
 
     if ";" in path_string:
-        fwd = Path(path_string.split(";")[0]).name
-        rev = Path(path_string.split(";")[1]).name
+        if path_string.count(";") == 1:
+            fwd = Path(path_string.split(";")[0]).name
+            rev = Path(path_string.split(";")[1]).name
+        else:  # three files per sample
+            fwd = Path(path_string.split(";")[1]).name
+            rev = Path(path_string.split(";")[2]).name
     else:
         fwd = Path(path_string).name
         rev = "NA"
@@ -252,13 +256,17 @@ def parse_to_mag(libraries):
         get_filename, orientation="rev"
     )
     libraries["short_reads_2"] = libraries["short_reads_2"].replace("NA", "")
+    libraries["short_reads_platform"] = libraries["instrument_model"].apply(get_sequencing_platform)
     libraries["longs_reads"] = ""
+    libraries["long_reads_platform"] = ""
     col2keep = [
         "archive_data_accession",
         "archive_sample_accession",
         "short_reads_1",
         "short_reads_2",
         "longs_reads",
+        "short_reads_platform",
+        "long_reads_platform",
     ]
     libraries = libraries[col2keep].rename(
         columns={
@@ -624,3 +632,30 @@ def is_merge_size_zero(
     if samples.shape[0] != 0 and library_selected.shape[0] == 0:
         return True
     return False
+
+
+def get_sequencing_platform(instrument_model: str) -> str:
+    """
+    Infer sequencing platform from the instrument model.
+
+    Args:
+        instrument_model (str): instrument model reported on ENA
+    Returns:
+        str: sequencing platform
+    """
+    if (instrument_model.startswith("Illumina") or
+            instrument_model.startswith("HiSeq") or
+            instrument_model.startswith("NextSeq")):
+        return "ILLUMINA"
+    elif instrument_model.startswith("454"):
+        return "454"
+    elif instrument_model.startswith("AB"):
+        return "AB"
+    elif instrument_model.startswith("BGISEQ"):
+        return "BGISEQ"
+    elif instrument_model == "Complete Genomics":
+        return instrument_model
+    elif instrument_model.startswith("Helicos"):
+        return "Helicos"
+    else:
+        return "Unknown"