Skip to content

Commit

Permalink
Update README.md
Browse files Browse the repository at this point in the history
  • Loading branch information
@rfd1
rfd1 committed Sep 2, 2022
1 parent 3cad02a commit 4e1ffae
Show file tree
Hide file tree
Showing 4 changed files with 338 additions and 336 deletions.
343 changes: 9 additions & 334 deletions README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -4,8 +4,12 @@ The purpose of this code is to enable reproduction
and facilitate extension of the computational
results associated with following work

DeJaco, R. F.; Majikes, J. M.; Liddle, J. A.; Kearsley, A. J.
Temperature-dependent Thermodynamic and Photophysical Properties of SYTO-13 Dye Bound to DNA.
DeJaco, R. F.; Majikes, J. M.; Liddle, J. A.; Kearsley, A. J.
Temperature-dependent Thermodynamic and Photophysical Properties of SYTO-13 Dye Bound to DNA.
[doi:10.1101/2202.08.31.505902](https://doi.org/10.1101/2022.08.31.505902).

For step-by-step instructions on reproducing the manuscript,
see [here](REPRODUCEME.md).

# Installation

Expand All @@ -16,340 +20,11 @@ The dependencies required can be installed via
pip install -r requirements.txt
```

# Reproducing the Manuscript

## Raw Data and Scaling

First, we read-in the data associated
with each data set (see Table 1 in paper)
and store as an instance of `src.get_data.RawData`.


>>> import sys, os; sys.path.append(os.getcwd())
>>> from src.get_data import RawData

The data for
each replicate plate
possessing single-stranded DNA with $D_k=1\times10^{-6}$ mol/L is
input into the following structures

>>> SS_A_1 = RawData(fluorescence_file_name="ssDNA_2-23-2021.xls", D_k=1e-6, t="SS", l="A")
>>> SS_B_1 = RawData(fluorescence_file_name="8-2-2021_GC_ssDNA.xls", D_k=1e-6, t="SS", l="B")
>>> SS_C_1 = RawData(fluorescence_file_name="1xSS_GC_11-3-2021_data.xls", D_k=1e-6, t="SS", l="C")

The data for
each replicate plate
possessing single-stranded DNA with $D_k=2\times10^{-6}$ mol/L is
input into the following structures

>>> SS_A_2 = RawData(fluorescence_file_name="2x_ssDNA_2-24-2021.xls", D_k=2e-6, t="SS", l="A")
>>> SS_B_2 = RawData(fluorescence_file_name="gc_2xssDNA_11-2-2021_data.xls", D_k=2e-6, t="SS", l="B")

The data for
each replicate plate
possessing double-stranded DNA with $D_k=1\times10^{-6}$ mol/L is
input into the following structures

>>> DS_A_1 = RawData(fluorescence_file_name="GC_0p5_dsDNA_11-2-2021.xls", D_k=1e-6, t="DS", l="A")
>>> DS_B_1 = RawData(fluorescence_file_name="gc_dsDNA_1uM_12-10-2021_data.xls", D_k=1e-6, t="DS", l="B")

The data for
each replicate plate
possessing double-stranded DNA with $D_k=2\times10^{-6}$ mol/L is
input into the following structures

>>> DS_A_2 = RawData(fluorescence_file_name="8-2-2021_GCdsDNA.xls", D_k=2e-6, t="DS", l="A")
>>> DS_B_2 = RawData(fluorescence_file_name="gc_dsDNA_2uM_12-9-2021_data.xls", D_k=2e-6, t="DS", l="B")

The data for the plate
without DNA
is input into the following structure

>>> A_1 = RawData(fluorescence_file_name="dyeOnly_11-6-2021_data.xls", D_k=0., t="None", l="A")

Having read-in the raw data, we plot it via

>>> from src.plot_raw_data import make_figure_2
>>> make_figure_2(SS_A_1, SS_B_1, SS_C_1, SS_A_2, SS_B_2, DS_A_1, DS_B_1, DS_A_2, DS_B_2, A_1)

which looks like

<p align="center">
<img
src="out/figure2.png"
width="250"
/>
</p>

We combine the replicate plates by storing them as a
`src.get_data.CombinedData` class.

>>> from src.get_data import CombinedData

The data for single-stranded DNA at $\mathbf{D}=1$ is

>>> SS_1 = CombinedData(SS_A_1, SS_B_1, SS_C_1)

The data for single-stranded DNA at $\mathbf{D}=2$ is

>>> SS_2 = CombinedData(SS_A_2, SS_B_2)

The data for double-stranded DNA at $\mathbf{D}=1$ is

>>> DS_1 = CombinedData(DS_A_1, DS_B_1)

The data for double-stranded DNA at $\mathbf{D}=2$ is

>>> DS_2 = CombinedData(DS_A_2, DS_B_2)


Having combined the data, we calculate $F_\mathrm{min}$ via

>>> F_min = 0
>>> from src.get_data import C_REF, F_REF
>>> for dataset in (SS_1, SS_2, DS_1, DS_2):
... wells = dataset.C*C_REF <= 0.5e-6
... "Num wells <= 0.5e-6 mol/L for %s, %i is %d" % (dataset.t, int(dataset.D), len(dataset.C[wells]))
... max_t_D = dataset.F[-1, wells].max()*F_REF
... if max_t_D > F_min:
... F_min = max_t_D
...
'Num wells <= 0.5e-6 mol/L for SS, 1 is 36'
'Num wells <= 0.5e-6 mol/L for SS, 2 is 24'
'Num wells <= 0.5e-6 mol/L for DS, 1 is 24'
'Num wells <= 0.5e-6 mol/L for DS, 2 is 24'


The value for $F_\mathrm{min}$ is

>>> F_min
231432.0

The subsets of the data are made via

>>> for dataset in (SS_1, SS_2, DS_1, DS_2):
... dataset.make_subset(F_min/F_REF)
... "Max temperature for %s, %i is %g K" % (dataset.t, int(dataset.D), dataset.T.max())
...
'Max temperature for SS, 1 is 316.5 K'
'Max temperature for SS, 2 is 324.5 K'
'Max temperature for DS, 1 is 322 K'
'Max temperature for DS, 2 is 329 K'

## Noise Removal

First, we import libraries used

>>> import numpy as np

For each dataset, compute $\mathbf{M}^\mathrm{LS}$ via Equation (21)
and store the results,

>>> from src.noise_removal import compute_M_LS
>>> F_hats = []
>>> for dataset in (SS_1, SS_2, DS_1, DS_2):
... M_LS = compute_M_LS(dataset.F, dataset.C)
... F_hats.append(np.outer(M_LS, dataset.C))
...

and then plot $\mathbf{F}$ vs $\widehat{\mathbf{F}}$ for
each combination via

>>> from src.plot_noise_removal import plot_Fhat_vs_F
>>> plot_Fhat_vs_F(
... (SS_1.F, SS_2.F, DS_1.F, DS_2.F),
... tuple(F_hats),
... (SS_1.T, SS_2.T, DS_1.T, DS_2.T),
... "figure3.png",
... sname=r"$\widehat{\mathbf{F}}_{ji}^\mathrm{LS}$")
...

This is Figure 3 in the main text, which looks like

<p align="center">
<img
src="out/figure3.png"
width="250"
/>
</p>

Subsequently, Equation (22) is solved
using `src.noise_removal.predictor_corrector`
and $V(\mathbf{M})$ and $V(\mathbf{C})$
are calulated


>>> from src.noise_removal import predictor_corrector
>>> RHO_SQUARED = 0.1
>>> for dataset in (SS_1, SS_2, DS_1, DS_2):
... dataset.M_tls, dataset.C_hat = predictor_corrector(
... dataset.F, dataset.C, RHO_SQUARED
... )
... dataset.Fhat_tls = np.outer(dataset.M_tls, dataset.C_hat.T)
...
... m, n = dataset.F.shape
... H = np.vstack([
... np.hstack([np.eye(n)*(RHO_SQUARED+np.inner(dataset.M_tls, dataset.M_tls)), np.zeros((n, m))]),
... np.hstack([np.zeros((m, n)), np.eye(m)*np.inner(dataset.C_hat, dataset.C_hat)])
... ])
... dF = dataset.Fhat_tls - dataset.F
... dC = dataset.C_hat - dataset.C
... f_star = (dF*dF).sum() + RHO_SQUARED*(dC*dC).sum()
... bbV = f_star / (m*(n-1))*np.linalg.inv(H)
... dataset.V_C = np.array([bbV[i, i] for i in range(n)])
... dataset.V_M = np.array([bbV[j, j] for j in range(n, n + m)])
...
... dataset.M_std = np.sqrt(dataset.V_M)
... dataset.C_std = np.sqrt(dataset.V_C)
...
Total number of iterations was 756
Total number of iterations was 1347
Total number of iterations was 1928
Total number of iterations was 3073

The results are plotted via Figure 4

>>> plot_Fhat_vs_F(
... (SS_1.F, SS_2.F, DS_1.F, DS_2.F),
... (SS_1.Fhat_tls, SS_2.Fhat_tls, DS_1.Fhat_tls, DS_2.Fhat_tls),
... (SS_1.T, SS_2.T, DS_1.T, DS_2.T),
... "figure4.png",
... sname=r"$\widehat{\mathbf{F}}_{ji}^\mathrm{TLS}$")
...

which looks like

<p align="center">
<img
src="out/figure4.png"
width="250"
/>
</p>

and Figure 5,

>>> from src.plot_noise_removal import plot_Chat_vs_C
>>> plot_Chat_vs_C(
... (SS_1.C, SS_2.C, DS_1.C, DS_2.C),
... (SS_1.C_hat, SS_2.C_hat, DS_1.C_hat, DS_2.C_hat),
... (SS_1.C_std, SS_2.C_std, DS_1.C_std, DS_2.C_std),
... "figure5.png"
... )
...

which looks like

<p align="center">
<img
src="out/figure5.png"
width="250"
/>
</p>


and Figure 6,

>>> from src.plot_noise_removal import plot_figure6
>>> plot_figure6(
... (SS_1.M_tls, SS_2.M_tls, DS_1.M_tls, DS_2.M_tls),
... (SS_1.M_std, SS_2.M_std, DS_1.M_std, DS_2.M_std),
... (SS_1.T, SS_2.T, DS_1.T, DS_2.T)
... )
...

which looks like

<p align="center">
<img
src="out/figure6.png"
width="250"
/>
</p>

## Parameter Extraction

First, we combine the DNA
concentrations associated with each DNA type
into an instance of `src.parameter_extraction.Parameters`

>>> from src.parameter_extraction import Parameters
>>> SS = Parameters(SS_1, SS_2)
>>> DS = Parameters(DS_1, DS_2)

These instances now
have methods that perform all parameter calculations;
we can readily plot the Figure 7 as

>>> from src.plot_params import plot_figure7
>>> plot_figure7(
... SS.T, SS.get_f(), SS.get_K(), SS.get_f_std(), SS.get_K_std(),
... DS.T, DS.get_f(), DS.get_K(), DS.get_f_std(), DS.get_K_std(),
... )
...

which looks like

<p align="center">
<img
src="out/figure7.png"
width="250"
/>
</p>

The association constant is calculated as

>>> K_a = DS.get_K()/2/C_REF*1e-6
>>> d_K_a = DS.get_K_std()/2/C_REF*1e-6
>>> "K_a at %3.2f K is %f +/- %f" % (DS.T[-24], K_a[-24], 3.*d_K_a[-24])
'K_a at 295.00 K is 0.671648 +/- 0.033107'

We plot Figure 8 via

>>> from src.plot_params import plot_figure8
>>> plot_figure8(SS, DS)
dg_SS at 295.00 K is -32.682584 +/- 0.088140
dg_DS at 295.00 K is -34.608172 +/- 0.108761



which looks like

<p align="center">
<img
src="out/figure8.png"
width="250"
/>
</p>


## Supplementary Figures

Figure S1 is made via

>>> from src.plot_raw_data import make_figure_S1
>>> make_figure_S1()

Figures S2, S3, S4 are made via

>>> from src.plot_params import plot_figure7, plot_figure8, plot_figure_S2, plot_figure_S3, plot_figure_S4, plot_figure_S5
>>> plot_figure_S2(SS_1, SS_2, DS_1, DS_2)
>>> plot_figure_S3(SS, DS)
>>> plot_figure_S4(SS, DS)

Figure S5 is made via

>>> from src.parameter_extraction import calculate_relative_brightness, calculate_relative_brightness_err
>>> rb = calculate_relative_brightness(SS.get_f(), DS.get_f())
>>> d_rb = calculate_relative_brightness_err(SS.M1, SS.M2, DS.M1, DS.M2,
... SS.V_M1, SS.V_M2, DS.V_M1, DS.V_M2)
>>> plot_figure_S5(SS.T, rb, d_rb)

# Data

The raw data can be found in the directory **data/**
The raw data can be found in the directory [data/](data/).

# Documentation

The documentation can be found in the **doc/** directory.
A pdf version of the documentation is available [here](doc/manual.pdf)

The documentation can be found in [doc/](doc/).
A pdf version of the documentation is available [here](doc/manual.pdf).
Loading

0 comments on commit 4e1ffae

Please sign in to comment.