Are quality filtered reads required to create your own error model?

[o-william-white]

Hello,

I would like to create my own error model. In the example below, should reads_R1.fastq.gz and reads_R2.fastq.gz be quality filtered prior to mapping?

Best wishes Ollie

bowtie2-build genomes.fasta genomes
bowtie2 -x genomes -1 reads_R1.fastq.gz -2 reads_R2.fastq.gz | \
samtools view -bS | samtools sort -o genomes.bam
samtools index genomes.bam

[HadrienG]

Hi! You should not do any quality filtering prior to create your own model (that is, if you want the same error profile than your raw reads). You should however ensure that all reads have the same length before mapping