Skip to content

Latest commit

 

History

History
263 lines (199 loc) · 10.9 KB

README.md

File metadata and controls

263 lines (199 loc) · 10.9 KB

The crosstalk between codon optimality and 3’ UTR cis-elements dictates mRNA stability

Santiago Gerardo Medina-Muñoz, Gopal Kushawah, Luciana Andrea Castellano, Michay Diez, Michelle Lynn DeVore, María José Blanco Salazar, Ariel A Bazzini 5/26/2020

Published data

A short description of each dataset accompanying this publication.

Table 1: zebrafish RNA-seq time course during MZT

RNA-seq gene level quantifications. Raw sequencing data have been deposited in NCBI Gene Expression Omnibus GSE148391. This table contains 35,117 rows and 31 columns. Each row represents a gene.

The column names contains all the relevant sample description:

  • Gene_ID -> zebrafish ensembl gene id
  • Treatment_x-y_hrs-RNAseq_z Here each column represents a single RNA-seq experiment containing the gene expression levels (Transcripts Per Million). The variables x, y, and z are placed holders: x: Some embryos were treated with alpha-amanitin to inhibit zygotic transcription. This takes only two values: aamanitin which denotes alpha-amanitin treated embryos and wt which represents untreated embryos. y: time post fertilization (hours). z: RNA-seq method ribo for ribosomal-RNA depletion and polya for poly-A selection.

A few rows and columns of the table are shown below:

Gene_ID Treatment_wt-0_hrs-RNAseq_ribo Treatment_wt-1_hrs-RNAseq_ribo
ENSDARG00000000002 2.71 2.20
ENSDARG00000000001 12.02 12.47
ENSDARG00000000019 49.06 39.72
ENSDARG00000000018 105.91 87.21
ENSDARG00000000069 219.59 163.14

Table 1

Table 2: mRNA stability during MZT

Messenger RNA degradation rates estimated from the alpha-amanitin time course. For additional information see the paper’s methods section: “Estimation of mRNA stability”.

95% confidence interval lower limit Columns description:

  • Gene_ID -> zebrafish ensembl gene id.
  • decay_rate -> estimated decay rate, negative values indicate unstable genes and positive values stable genes.
  • std.error -> standard error of the estimate.
  • conf.low -> lower limit 95% confidence interval.
  • conf.high -> upper limit 95% confidence interval.

A few rows of the table are shown below:

Gene_ID decay_rate std.error conf.low conf.high
ENSDARG00000000001 -0.4317706 0.0610166 -0.5677241 -0.2958171
ENSDARG00000000018 0.1329993 0.0487225 0.0244389 0.2415597
ENSDARG00000000019 0.0667906 0.0190031 0.0244490 0.1091321
ENSDARG00000000068 0.0057338 0.0202691 -0.0394285 0.0508961
ENSDARG00000000069 -0.3366622 0.0318811 -0.4076978 -0.2656266
ENSDARG00000000086 0.3106115 0.0428094 0.2152261 0.4059968

Table 2

Table 3: Training/testing data to train machine learning predictor of mRNA stability

Data that was used to train and evaluate machine learning preditor model. This table contains 75,351 rows and 8 columns.

Columns description:

  • gene_id -> ensembl gene id.
  • specie -> specie vertebrate (human = H. sapiens, fish = D. rerio, mouse = M. musculus, and xenopus = X. tropicalis).
  • cell_type -> cell type from where mRNA stability measurements were derived.
  • datatype -> How where mRNA stability mesurements generated?:
    • endogenous Actinomycin D was used to block transcription
    • aamanitin ribo embryos treated with alpha-amanitin (RNA-seq ribosomal-RNA depletion)
    • aamanitin polya embryos treated with alpha-amanitin (RNA-seq poly-A selection)
    • slam-seq SLAM-seq.
  • decay_rate -> mRNA degradation rates (see the note below).
  • utrlenlog -> 3’ UTR length log-transformed.
  • cdslenlog -> cds length log-transformed.
  • allocation -> this variable denotes wether the given observations was used for training or for testing (validation). Note: the same gene-id is never in training or testing simultaneously.

Notes:

  • The mRNA degradation rates in this table were standardized (mean = 0 and standard deviation = 1). The next table below shows the mean and standard deviations of the original data. By applying the inverse transform, the values in the original scale can be recovered.
  • This table is missing the codon composition (codon frequencies). Codons frequencies were computed, for each gene, from the longest coding isoform sequence.
specie cell_type datatype mean_decayrate stdeviation_decayrate
fish embryo mzt aamanitin polya -0.0629208 0.1860839
fish embryo mzt aamanitin ribo -0.0072156 0.0032341
human 293t endogenous 0.0051652 0.0432408
human hela endogenous -0.0009781 0.0542492
human k562 endogenous -0.0176985 0.0668504
human k562 slam-seq -0.1313344 0.0672748
human RPE endogenous -0.0071829 0.0597563
mouse mES cells slam-seq -0.1961037 0.0853733
xenopus embryo mzt aamanitin ribo -0.0016150 0.0007771

means and strandar deviations

Decay rates, for the given specie, were obtained from the following publications:

  • zebrafish: This study; AA Bazzini, F del Viso, MA Moreno‐Mateos… - The EMBO journal, 2016.
  • human: Q Wu, SG Medina, G Kushawah, ML DeVore… - Elife, 2019.
  • xenopus: AA Bazzini, F del Viso, MA Moreno‐Mateos… - The EMBO journal, 2016.
  • mouse: VA Herzog, B Reichholf, T Neumann, P Rescheneder… - Nature …, 2017.

A few rows of the table are shown below excluding the coding column:

gene_id specie cell_type datatype decay_rate utrlenlog cdslenlog allocation
ENSG00000149428 human 293t endogenous 0.9591786 7.705713 8.007700 training
ENSG00000183826 human 293t endogenous -0.2203851 8.784928 7.517521 training
ENSDARG00000007181 fish embryo mzt aamanitin ribo 0.6516244 5.236442 7.165494 training
ENSDARG00000025820 fish embryo mzt aamanitin ribo -2.1045739 5.056246 7.321850 training
ENSDARG00000017244 fish embryo mzt aamanitin polya 1.5186344 8.373323 9.088625 training

Table 3

Table 4: Predictions and residual values during MZT for zebrafish and xenopus.

This data is part of Fig. 2 (see paper). This table contains the codon predicted stability for zebrafish and xenopus during MZT. This table contains 10,668 rows and 4 columns.

Columns description:

  • gene_id ->: ensembl gene id, the genes here are maternally deposited.
  • specie ->: either zebrafish or xenopus.
  • residual ->: residual values, values close to zero indicate that the model predicts well the stability and large values that the model is far from the observed value. The residual is the difference between observed and predicted mRNA stability.

A few rows of the table are shown below:

gene_id specie predicted residual
ENSXETG00000017158 xenopus 0.0736592 0.4100507
ENSDARG00000041113 fish -0.0636230 0.6425244
ENSXETG00000019722 xenopus -0.6945703 -0.0959954
ENSXETG00000020444 xenopus 0.4641538 0.7438621
ENSXETG00000013779 xenopus 0.1119481 -2.0316289

Table 4

Table 5: Gene level mesurements of codon optimality.

See the methods section “Measuring codon optimality at the gene level” and Additional file Figure S3. This table contains numerical measurements of codon optimality in some endogenous genes, we have generated two such measurements PLS1 and PLS2, positive values are associated with enrichment in optimal codons (stabilizing codon) and negative values with enrichment in non-optimal codons (destabilizing codon).

This table contains 57,627 rows and 4 columns.

Columns description:

  • gene_id ->: ensembl gene id.
  • PLS1 -> measurement 1 of codon optimality
  • PLS2 -> measurement 2 of codon optimality
  • specie -> vertebrate (human, mouse, xenopus, or zebrafish).
gene_id PLS1 PLS2 specie
ENSDARG00000099464 0.7106535 -2.4052864 fish
ENSG00000181396 2.6401594 -1.4055193 human
ENSG00000123576 0.6598256 -1.5668842 human
ENSMUSG00000040034 -3.7472500 2.0859899 mouse
ENSDARG00000101190 -1.1129394 0.7257975 fish

Table 5

Table 6: Reporter sequences

This table contains 4 rows and 2 columns (Fig. 4e).

Columns description:

  • sequence_id ->: The sequence id.
  • PLS1 -> The reporter DNA sequence.

Next, you can find the first column of this table.

sequence_id
CODING-optimal
CODING-non_optimal
3UTR-mir17
3UTR-mir17_mutant

Table 6

Table 7: Codon frequencies of endogenous genes used to train machine learning model

This table contains the codon frequencies of the endogenous genes for zebrafish, Xenopus, mouse, and human. The frequencies were determined from the longest coding sequence isoform for each transcript.

Together this table and Table 3 can be used to train the machine learning model to predict mRNA stability.

Columns description:

  • gene_id ->: ensembl gene id.
  • AAA -> frequency in transcript for codon AAA.
  • AAC -> frequencies in transcript for codon AAC.
  • etc.
gene_id AAA AAC AAG AAT
ENSDARG00000079869 5 7 12 3
ENSG00000179604 1 0 17 5
ENSG00000169813 6 8 10 3
ENSG00000167700 1 1 7 2
ENSG00000083454 7 16 22 4

Table 7