# Translate all reads in a .fastq into protein sequence from deep mutational scanning experiment

I'm working with paired-end NGS reads from an Illumina platform. The sequences I have are all of the same gene, but have one or more substitution mutations each.

Here is a rough workflow for generating the reads:

1. Expression of a library of sequence variants for a heterogeneous protein of interest in phage.
2. Selection of high-fitness variants according to a custom assay.
3. Extraction of the variant pool via amplification of the variable region of the gene of interest.
4. Preparation of Illumina sequencing library from the amplicon pool.
5. Deep sequencing of the amplicon pool.

I've already combined the paired-end reads with USearch and manually removed sequences with subpar quality scores, but I now need to translate all of the remaining DNA sequences into amino acid sequences. I'm uncertain of where to begin addressing this issue, and I was hoping someone could point me in the direction of an application or python package that could assist me.

My goal: Count the abundances of each sequence variant within the Illumina data to identify which of the protein sequences are the most fit.

• Won't you need to map to a reference? You will need to know what frame to use to translate the reads. Unclear to me what the experimental design is here. Where do these reads come from? Is it an amplicon library? (If it is amplicon it is possible that you don't need to map anywhere.) Aug 30, 2021 at 17:23
• Reads are coming from a mixed population of phages which express multiple sequence variants of a protein of interest on their surface. The protein of interest is then put under selection. after selection the gene sequences of phages were collected and quantified to identify copy number of each variant in the population. This allows us to determine the fitness of each variant by seeing how many of that sequence there is. I suppose I might need a reference map, but I am pretty inexperienced to know which program would be best.
– Paul
Aug 30, 2021 at 17:41
• "gene sequences of phages were collected" how? Presumably by sequencing, but what was the sequencing library preparation procedure? Was it whole genome shotgun e.g. nextera prep, locus-specific amplification (i.e. your protein of interest), barcode amplicon, something else? Presumably you limited sequencing to your protein of interest somehow, otherwise this would be a somewhat inefficient workflow. Aug 30, 2021 at 18:44
• Thank you for these questions! This is not my wet experiment, I am simply trying to use the data. But I can tell you the read libraries were generated with locus specific PCR.
– Paul
Aug 30, 2021 at 20:33
Based on back and forth in the comments, I think that you might be interested in a deep mutational scanning workflow, for example dms_tools2 or enrich2.