Is it possible to predict how proteins coded from mRNA will be cleaved?
The reason I was interested in this is because I did some initial work to translate the raw Coronavirus RNA sequences, which you can see here: https://github.com/imranq/coronavirus. The background section below has more information on my approach.
With the "naive" approach to translate coronavirus RNA, using AUG as the start codon, I am able to identify 7 out of 24 proteins from Zhang lab sequences. However, I also did some string matching between my amino acid sequences and the Zhang lab final amino acid sequences. After doing that comparison, I see I have 21 out of 24 protein sequences coded in uncleaved protein sequences.
It would be nice to get to the final 24 proteins straight from the mRNA strand. Please help. Thanks!
Here is the series of steps I used to convert coronavirus RNA into proteins.
1. The raw RNA sequences are here (sourced from the NIH): https://github.com/imranq/coronavirus/blob/main/data/rawrna.json 2. The script to translate the RNA is here (assumes AUG is the start codon): https://github.com/imranq/coronavirus/blob/main/scripts/translate.js The translated protein sequences are here: https://github.com/imranq/coronavirus/blob/main/data/processed/translatedProteins.json 3. The script used to generate a comparison between the translated proteins and the known protein sequences is here (using levenshtein distance as our difference metric): https://github.com/imranq/coronavirus/blob/main/scripts/compareTranslated.js The data generated from that script is here https://github.com/imranq/coronavirus/blob/main/data/processed/translatedComparison.json 4. Then we detect cleavage in proteins by running a sliding window across RNA segments between the known and translated proteins. The script to do so is here: https://github.com/imranq/coronavirus/blob/main/scripts/detectCleavage.js 4. Finally we merge the datasets we generated from translation and protein complexes into one dataset, translated proteins and protein complexes matched to known proteins. The data generated from this algorithm is located here https://github.com/imranq/coronavirus/blob/main/data/processed/mergedProteins.json
At the end of this pipeline, we get 21 out of the 24 proteins matched straight from RNA.
Here are the papers / articles I researched to answer this question
Zhan Lab - SARS CoV 2 - Nucleotide, Coding Region, and Protein sequences https://zhanglab.ccmb.med.umich.edu/COVID-19/
More in the readme here: https://github.com/imranq/coronavirus/blob/main/README.MD