StressTranslatome

Legend

Gene/Label: User provided Gene name or Label of 5' UTR sequence.
Score Formula: Score = e^Prescore / (1 + e^Prescore)
Prescore = β₀ + β₁·A + β₂·B + β₃·C + β₄·D + β₅·A·B + β₆·A·C + β₇·A·D + β₈·B·C + β₉·B·D + β₁₀·C·D
Where: A = uORFs, B = Atf4like, C = Length, D = %GC
ORFs: a discrete variable that counts all possible uORFs starting by AUG.
Atf4like: a categorical variable (0, 1, 2, or 3) that counts the reading frames in which we found a uORF that has not yet encountered a STOP codon and was preceded by at least one already terminated uORF (it was not necessary for the already terminated uORF to be in the same reading frame).
Length: a continuous variable that counts the number of nucleotides present in the 5′UTR.
%GC: a continuous variable that represents the percentage of G or C nucleotides relative to the total length.
Condition for inclusion in results: Only positive hits are shown. We defined positive hits as those having a Score higher than 0.7 and at least one uORF.

About StressTranslatome

Understanding the intricate mechanisms governing gene expression regulation is crucial for deciphering neuronal responses to cellular stress at the physiological (i.e. synaptogenesis) and pathophysiological level (i.e. neurodegenerative diseases). These rapid adaptive changes depend on the translation of specific proteins with specialized 5′ untranslated regions (5′UTRs), triggered by the phosphorylation of eukaryotic initiation factor 2 alpha (eIF2α), while normal cellular translation remains largely inhibited. StressTranslatome is a highly specific tool designed to identify mRNAs susceptible to be regulated by p-eIF2α. We compiled a database of 5′UTRs using Ensembl canonical transcripts from the GRCh38.p14 genome build. Ensembl IDs were used to extract coding sequences and cDNA via the REST API, and 5'UTR regions were identified. We applied translation efficiency-based filters on existing databases for p-eIF2α-dependent translation to obtain reliable training and tasting datasets. A multiple logistic regression model predicted scores for p-eIF2α-driven translation, using 5'UTR length, GC percentage, predicted number of upstream open reading frames (uORFs) starting with AUG, and the count of reading frames with common features of Atf4 as a reference. Within the framework of this webtool, a user may submit 5′UTRs for the filtration of sequences predicted to have positive p-eIF2α-driven translation, indicated by a score higher than 0.7 and having at least one uORF. A manuscript accompanying the development of this tool along with further analysis has been submitted for review and will be linked upon publication.

The tool accepts either direct input of a gene and 5′UTR sequence, or batch analysis via multi-sequence FASTA file upload. Results above a predictive threshold are displayed and available for CSV download.

Funding

This work was supported by the Spanish Ministry of Science and Innovation and Agencia Estatal de Investigación plus FEDER Funds through grants PID2023-149767OB-I00/MICIUN/AEI/10.13039/501100011033/FEDER-UE (FJM), PID2023-150068OB-I00 (BO) and PID2022-136511OB-I00 (RV). This work was also funded by the Spanish Institute of Health Carlos III by project reference AC20/00009 -FEDER/UE and ERANET ERA-CVD_JTC2020-015; and, “María de Maeztu Programme” for Units of Excellence in R&D (award CEX2018-000792-M). This project was funded in part by TUBITAK Research GrantNo: 220N252 (AG).