fixed compression of basecalled data, added reference alignment step

workflow/rules/baseline_pipeline.smk

@@ -52,7 +52,7 @@ rule basecall_pod5:
         model="hac|fast"
     shell:
         """
-        dorado basecaller --models-directory {config[dorado_model_dir]} --emit-fastq {params.benchmarking} {params.min_qscore} {params.dorado_model} {input.pod5} > {output}
+        dorado basecaller --models-directory {config[dorado_model_dir]} --emit-fastq {params.benchmarking} {params.min_qscore} {params.dorado_model} {input.pod5} | gzip -c > {output}
         """
 
 rule concatenate_basecalled_fastq:
@@ -66,4 +66,32 @@ rule concatenate_basecalled_fastq:
     shell:
         """
         zcat {input} > {output}
+        """
+
+rule align_reads_to_reference:
+    input:
+        fastq='../data/basecalled_reads/{model}.fastq.gz',
+        ref='../data/reference_genomes/full_reference.mmi'
+    output:
+        '../data/aligned_reads/{model}_to_genome.sam'
+    threads: 32
+    conda: 
+        '../envs/minimap.yaml'
+    shell:
+        """
+        minimap2 -ax map-ont -t {threads} {input.ref} {input.fastq} > {output}
+        """
+
+rule convert_sam_to_bam:
+    input:
+        '../data/aligned_reads/{model}_to_genome.sam'
+    output:
+        '../data/aligned_reads/{model}_to_genome.sorted.bam'
+    threads: 32
+    conda:
+        '../envs/samtools.yaml'
+    shell:
+        """
+        samtools view -b -@ {threads} {input} | samtools sort -@ {threads} > {output}
+        samtools index -@ {threads} {output}
         """