luciferase 2.2.3

Helper functions for luciferase data

luciferase

Tool for plotting luciferase reporter data. Thanks due to Joshua Chiou and Mei-lin Okino for inspiration and contributions.

Installation

pip3 install luciferase

or

pip3 install --user luciferase

Command-line interface for barplots

Barplots of enhancer activity

A command-line tool called luciferase-barplot for creating bar plots is included. After installing luciferase, you can use it like this:

luciferase-barplot example.json example.pdf
luciferase-barplot example.csv example.png
luciferase-barplot example.tsv example.svg
luciferase-barplot example.xls example.pdf
luciferase-barplot example.xlsx example.png

JSON, CSV, TSV, or Excel files may be used as inputs, and output can be written in PDF, PNG, or SVG.

See also the help message:

luciferase-barplot -h

Examples of luciferase reporter data in JSON format:

{
  "Non-risk, Fwd": [8.354, 12.725, 8.506],
  "Risk, Fwd": [5.078, 5.038, 5.661],
  "Non-risk, Rev": [9.564, 9.692, 12.622],
  "Risk, Rev": [10.777, 11.389, 10.598],
  "Empty": [1.042, 0.92, 1.042]
}

{
  "Alt, MIN6": [5.47, 7.17, 6.15],
  "Ref, MIN6": [3.16, 3.04, 4.34],
  "Empty, MIN6": [1.07, 0.83, 0.76],
  "Alt, ALPHA-TC6": [2.50, 3.47, 3.33],
  "Ref, ALPHA-TC6": [2.01, 1.96, 2.31],
  "Empty, ALPHA-TC6": [1.042, 0.92, 1.042]
}

Examples of luciferase reporter data in TSV format:

Non-risk, Fwd	Risk, Fwd	Non-risk, Rev	Risk, Rev	Empty
8.354	5.078	9.564	10.777	1.042
12.725	5.038	9.692	11.389	0.92
8.506	5.661	12.622	10.598	1.042

Ref, untreated	Alt, untreated	Empty, untreated	Ref, dex	Alt, dex	Empty, dex
33.2	19.7	1.0	149.4	44.6	1.1
30.3	16.2	1.0	99.7	37.6	1.0
33.3	18.3	1.0	124.5	37.7	0.9

Significance indicators will be written above the bars: *** if p<0.001, ** if p<0.01, * if p<0.05, ns otherwise.

Here are the resulting plots:

Barplots of allelic ratio

A second tool called luciferase-ratioplot takes the same input data and produces a comparative plot of allelic ratios:

luciferase-ratioplot --xlab control dexamethasone --ylab "Ref:Alt ratio" --title Default ratio.json ratio.png
luciferase-ratioplot --xlab control dexamethasone --ylab "Alt:Ref ratio" --title Inverted --invert ratio.json ratio.png

The resulting plot shows the estimated allelic ratio of enhancer activity with confidence intervals (95% by default). Here is an example input dataset and plot:

JSON:

{
  "Ref, untreated": [33.2, 30.3, 33.3],
  "Alt, untreated": [19.7, 16.2, 18.3],
  "Empty, untreated": [1.0, 1.0, 1.0],
  "Ref, dex": [149.4, 99.7, 124.5],
  "Alt, dex": [44.6, 37.6, 37.7],
  "Empty, dex": [1.1, 1.0, 0.9]
}

TSV:

Alt, dex	Ref, dex	Empty, dex	Alt, untreated	Ref, untreated	Empty, untreated
44.6	149.4	1.1	19.7	33.2	1.0
37.6	99.7	1.0	16.2	30.3	1.0
37.7	124.5	0.9	18.3	33.3	1.0

Meta-analysis

For this section, we'll use another included command called luciferase-swarmplot. It functions exactly like luciferase-barplot except that individual data points will be plotted over the bars.

It may be that we have performed two or more experiments (from separate minipreps) and wish to meta-analyze the results. As an example, let's consider the results of two identical experiments on a regulatory variant at the SIX3 locus: SIX3-MP0 and SIX3-MP1. First we'll plot both datasets separately:

luciferase-swarmplot six3-mp0.json six3-mp0.png --light-colors '#DECBE4' '#FED9A6' '#FBB4AE' --dark-colors '#984EA3' '#FF7F00' '#E41A1C' --title 'SIX3-MP0'
luciferase-swarmplot six3-mp1.json six3-mp1.png --light-colors '#DECBE4' '#FED9A6' '#FBB4AE' --dark-colors '#984EA3' '#FF7F00' '#E41A1C' --title 'SIX3-MP1'

We can see that the results are fairly consistent in character, but checking the y-axis tells us that they are on different scales. Intuitively, we might conclude from these results that there are allelic effects under all three conditions. Ideally though, we would like to use all of the data at once for one plot to get the most accurate conclusions about allelic effects.

We might simply combine the data into one dataset, (as here) and plot it:

luciferase-swarmplot six3-meta-nobatch.json six3-meta-nobatch.png --light-colors '#DECBE4' '#FED9A6' '#FBB4AE' --dark-colors '#984EA3' '#FF7F00' '#E41A1C'

The bar heights look reasonable, and the allelic effects appear clear from looking at them, but all of the hypothesis tests returned non-significant results. What gives?

The answer is that combining data from experiments with different scales breaks the assumptions of the significance test (a t-test). To meta-analyze these data in a useful way, we first need to re-normalize the two experiments to put both of them on the same scale. luciferase-barplot and luciferase-swarmplot will re-normalize the data automatically if the dataset includes an additional entry ("Batch") indicating the batch of each data point, as in this example: SIX3-META.

{
  "Alt, untreated": [19.7, 16.2, 18.3, 6.5, 8.0, 4.4],
  "Ref, untreated": [33.2, 30.3, 33.3, 8.4, 13.6, 17.1],
  "Empty, untreated": [1.0, 1.0, 1.0, 1.1, 1.0, 0.9],
  "Alt, hi_cyt_noTNFA": [11.0, 8.8, 10.1, 3.2, 3.7, 3.3],
  "Ref, hi_cyt_noTNFA": [17.1, 16.7, 18.8, 7.6, 6.7, 5.5],
  "Empty, hi_cyt_noTNFA": [1.1, 0.9, 1.0, 1.1, 0.9, 1.0],
  "Alt, hi_cyt": [10.8, 10.9, 9.1, 3.1, 2.7, 4.0],
  "Ref, hi_cyt": [17.8, 16.1, 18.0, 7.7, 7.0, 7.1],
  "Empty, hi_cyt": [1.0, 1.0, 1.0, 1.0, 1.0, 1.1],
  "Batch": [0, 0, 0, 1, 1, 1]
}

Here is what the results look like when they're re-normalized to correct for batch

luciferase-swarmplot six3-meta.json six3-meta.png --light-colors '#DECBE4' '#FED9A6' '#FBB4AE' --dark-colors '#984EA3' '#FF7F00' '#E41A1C' --title 'SIX3-META'

See a more detailed explanation of the normalization procedure here