Fix: Update src/create_fasta.py

In 15638ff, binsplitting was refactored. This change removed the minsize keyword
from vambtools.write_bins, which src/create_fasta.py relies on, breaking the
script.
Fix this, by making the following changes:
* Read in the FASTA in two passes, one to figure out which sequences to store in
  memory, and one which stored them
* Filter the bins in the script before calling write_bins
* Change the default max bins from the overly conservative 250 to 1000.

src/create_fasta.py

@@ -1,6 +1,7 @@
 import sys
 import argparse
 import vamb
+import pathlib
 
 parser = argparse.ArgumentParser(
     description="""Command-line bin creator.
@@ -11,7 +12,7 @@
 
 parser.add_argument("fastapath", help="Path to FASTA file")
 parser.add_argument("clusterspath", help="Path to clusters.tsv")
-parser.add_argument("minsize", help="Minimum size of bin", type=int, default=0)
+parser.add_argument("minsize", help="Minimum size of bin in bp", type=int, default=0)
 parser.add_argument("outdir", help="Directory to create")
 
 if len(sys.argv) == 1:
@@ -20,10 +21,21 @@
 
 args = parser.parse_args()
 
+# Read in FASTA files only to get its length. This way, we can avoid storing
+# in memory contigs for sequences that will never get output anyway
+lens: dict[str, int] = dict()
+with vamb.vambtools.Reader(args.fastapath) as file:
+    for record in vamb.vambtools.byte_iterfasta(file):
+        lens[record.identifier] = len(record)
+
 with open(args.clusterspath) as file:
     clusters = vamb.vambtools.read_clusters(file)
 
+clusters = {
+    cluster: contigs
+    for (cluster, contigs) in clusters.items()
+    if sum(lens[c] for c in contigs) >= args.minsize
+}
+
 with vamb.vambtools.Reader(args.fastapath) as file:
-    vamb.vambtools.write_bins(
-        args.outdir, clusters, file, maxbins=None, minsize=args.minsize
-    )
+    vamb.vambtools.write_bins(pathlib.Path(args.outdir), clusters, file, maxbins=None)

vamb/vambtools.py

@@ -617,15 +617,15 @@ def write_bins(
     directory: Path,
     bins: dict[str, set[str]],
     fastaio: Iterable[bytes],
-    maxbins: Optional[int] = 250,
+    maxbins: Optional[int] = 1000,
 ):
     """Writes bins as FASTA files in a directory, one file per bin.
 
     Inputs:
         directory: Directory to create or put files in
         bins: dict[str: set[str]] (can be loaded from clusters.tsv using vamb.cluster.read_clusters)
         fastaio: bytes iterator containing FASTA file with all sequences
-        maxbins: None or else raise an error if trying to make more bins than this [250]
+        maxbins: None or else raise an error if trying to make more bins than this [1000]
     Output: None
     """