Increase base memory (#112)

* revert: import typing.Literal

* chore: remove unused `cfg.temp_root` directory

* chore: increase trimming and contaminant screening memory request

* chore: if contaminant genomes already donwloaded, touch output to make sure Snakemake sees it exists

Author: JoshLoecker

Snakefile

@@ -1,4 +1,5 @@
 import sys
+from typing import Literal
 
 from utils.parse import Config, SampleData, print_key_value_table
 
@@ -14,7 +15,6 @@ onstart:
             [
                 ("Samples", cfg.sample_filepath),
                 ("Data", cfg.data_root),
-                ("Temporary", cfg.temp_root),
                 ("COMO", cfg.como_root),
                 ("Logging", cfg.logs_root),
                 ("Genome", cfg.genome.species_dir),
@@ -227,6 +227,7 @@ rule download_contaminant_genomes:
                 expected_files="${{expected_files[index]}}"
                 if [ $existing_files -eq $expected_files ]; then
                     echo "[fastq_screen] Skipping genome download for '$genome' because it is already present at '$outdir'" >> {log}
+                    touch "$outdir" "$outdir/*"
                     continue
                 fi
             fi
@@ -476,7 +477,7 @@ rule trim_paired:
         r2_fastq=f"{cfg.data_root}/{{tissue}}/trim/{{tissue}}_{{tag}}_2.fastq.gz",
         r2_report=f"{cfg.data_root}/{{tissue}}/trim/{{tissue}}_{{tag}}_2_trimming_report.txt",
     resources:
-        mem_mb=lambda wildcards, attempt: 1024 * attempt,
+        mem_mb=lambda wildcards, attempt: 4096 * attempt,
         runtime=lambda wildcards, attempt: 45 * attempt,
         tissue=lambda wildcards: wildcards.tissue,
     threads: 4
@@ -771,7 +772,7 @@ rule contaminant_screen_paired:
     params:
         output_dir=f"{cfg.data_root}/{{tissue}}/fq_screen",
     resources:
-        mem_mb=lambda wildcards, attempt: 1024 * attempt,
+        mem_mb=lambda wildcards, attempt: 4069 * attempt,
         runtime=lambda wildcards, attempt: 10 * attempt,
         tissue=lambda wildcards: wildcards.tissue,
     threads: 5
@@ -824,7 +825,7 @@ rule contaminant_screen_single:
     params:
         output_dir=f"{cfg.data_root}/{{tissue}}/fq_screen"
     resources:
-        mem_mb=lambda wildcards, attempt: 1024 * attempt,
+        mem_mb=lambda wildcards, attempt: 4096 * attempt,
         runtime=lambda wildcards, attempt: 20 * attempt,
         tissue=lambda wildcards: wildcards.tissue,
     threads: 5

utils/parse.py

@@ -103,7 +103,6 @@ class Config:
     sample_filepath: Path
     root: Path
     data_root: Path
-    temp_root: Path
     como_root: Path
     logs_root: Path
     experiment_name: str
@@ -163,20 +162,13 @@ def create(cls, config: dict[str, Any]) -> "Config":
         taxon_id: int = int(config["GENOME"]["TAXONOMY_ID"])
         species_name: str = species_from_taxon(taxon_id=taxon_id)
 
-        sample_filepath = Path(config["MASTER_CONTROL"])
-        data_root = Path(root, experiment_name, "data")
-        temp_root = Path(root, experiment_name, "temp")
-        como_root = Path("COMO_input", experiment_name)
-        logs_root = Path(config["LOG_DIR"], experiment_name)
-        species_dir = Path(config["GENOME"]["SAVE_DIR"], species_name)
 
         return cls(
             sample_filepath=Path(config["MASTER_CONTROL"]),
             root=root,
-            data_root=data_root,
-            temp_root=temp_root,
-            como_root=como_root,
-            logs_root=logs_root,
+            data_root=Path(root, experiment_name),
+            como_root=Path("COMO_input", experiment_name),
+            logs_root=Path(config["LOG_DIR"], experiment_name),
             experiment_name=experiment_name,
             local_fastq_filepath=Path(fastq_files) if fastq_files else None,
             species_name=species_name,
@@ -195,8 +187,8 @@ def create(cls, config: dict[str, Any]) -> "Config":
                 bypass_genome_validation=config["BYPASS_GENOME_VALIDATION"],
             ),
             genome=Genome(
-                species_dir=species_dir,
-                contaminants_dir=Path(config["ROOTDIR"], "FastQ_Screen_Genomes"),
+                species_dir=Path(config["GENOME"]["SAVE_DIR"], species_name),
+                contaminants_dir=Path(config["ROOTDIR"], "fastq_screen_genome"),
                 taxon_id=taxon_id,
                 version=config["GENOME"]["VERSION"],
                 type=config["GENOME"]["FASTA_TYPE"],