# Spectrophores
[](https://doi.org/10.1186/s13321-018-0268-9) [](https://github.com/silicos-it/spectrophore/actions/workflows/test.yml)
This module contains python code to calculate `spectrophores` from molecules. It is uses `RDKit` to process molecules, and `numba` to improve the performance.
The technology and its applications have been described in [*Journal of Cheminformatics* (2018) **10**, 9](https://jcheminf.biomedcentral.com/articles/10.1186/s13321-018-0268-9). The paper is also included in this distribution.
The `spectrophore` code can be used in two ways:
- As a standalone program to convert the molecules in a sd-file into their corresponding `spectrophores`
- As a `python` module to import in your own `python` code.
In the following sections, both usages will be documented.
This repository also includes `spectrophore_notebook.ipynb`, a portable Jupyter Notebook which could be useful for quick and interactive data analysis with spectrophores.
## Installation
### Installation using pip (recommended)
1. Create and activate a new Anaconda environment
```
conda create --name spectrophore python=3
```
```
conda activate spectrophore
```
2. Install the repo via pip. The RDKit, Numba and other dependencies should be installed automatically.
```
pip install uamc-spectrophore
```
### Installation from Source
1. Clone the repository and navigate to it
```
git clone https://github.com/silicos-it/spectrophore.git
```
```
cd spectrophore
```
2. Create and activate a new Anaconda environment
```
conda create --name spectrophore python=3
```
```
conda activate spectrophore
```
3. Install the repo by using pip. The RDKit, Numba and other dependencies should be installed automatically.
```
pip install .
```
#### For Contributors
To install with additional development dependencies (`pytest`, `ruff`, etc.), use:
```
pip install .[develop]
```
## Usage
### 1. Usage as a standalone program
After installation, one should be able to use the `spectrophore` package as a standalone program to calculate `spectrophores` from a sd-file with molecules.
```console
> python -m spectrophore -h
```
or even more convenient:
```console
> spectrophore -h
```
This will provide you with all details on how to calculate `spectrophores` from a sd-file, as well as all the configurable settings:
```console
usage: spectrophore [-h] [-n {none,mean,all,std}] [-m {classic,full}] [-s {none,unique,mirror,all}] [-a {1,2,3,4,5,6,9,10,12,15,18,20,30,36,45,60,90,180}] [-r RESOLUTION] [-p MAX_WORKERS]
[--silent] -i INFILE -o OUTFILE
Convert a SDF-file containing 3D structures to a csv with the corresponding spectrophores fingerprints
options:
-h, --help show this help message and exit
-n {none,mean,all,std}, --norm {none,mean,all,std}
Normalization setting (default: all)
-m {classic,full}, --mode {classic,full}
Which type of spectrophore to calculate (default: classic)
-s {none,unique,mirror,all}, --stereo {none,unique,mirror,all}
Stereo setting (default: none)
-a {1,2,3,4,5,6,9,10,12,15,18,20,30,36,45,60,90,180}, --accuracy {1,2,3,4,5,6,9,10,12,15,18,20,30,36,45,60,90,180}
Accuracy setting (default: 20)
-r RESOLUTION, --resolution RESOLUTION
Resolution setting (>0) (default: 3)
-p MAX_WORKERS, --np MAX_WORKERS
Number of processors to use; -1 is all processors (default: -1)
--silent Don't show a progressbar (default: False)
required arguments:
-i INFILE, --in INFILE
Input sdf file (default: None)
-o OUTFILE, --out OUTFILE
Output spectrophore file (default: None)
```
### 2. Usage as a python module
### 2.1. Introduction
Once you have installed the `uamc-spectrophore` package, you are ready to calculate `spectrophores`. In its most simple form, `spectrophores` can be computed as follows:
```python
>>> from spectrophore import spectrophore
>>> from rdkit import Chem
>>> from rdkit.Chem import AllChem
>>> mol = Chem.MolFromSmiles("c1ncncc1")
>>> mol = Chem.AddHs(mol)
>>> AllChem.EmbedMolecule(mol, randomSeed=1)
0
>>> calculator = spectrophore.SpectrophoreCalculator(normalization='none')
Probes initialised: 48 number of probes in total
12 probes are used due to the imposed stereo flag
>>> calculator.calculate(mol)
array([ 1.4826062, 2.0534244, 1.6182035, 3.1277552, 2.5039089,
5.157424 , 2.3022916, 1.6652936, 3.4908977, 4.0566797,
5.020808 , 3.0757458, 0.6589167, 0.7786141, 4.324721 ,
4.3273506, 2.928256 , 4.132863 , 1.9475778, 3.5233636,
4.6044445, 4.1554613, 4.5122004, 5.2647495, 46.877754 ,
71.05832 , 96.21371 , 94.29615 , 51.998947 , 63.523525 ,
24.393661 , 80.728264 , 73.68167 , 65.27555 , 56.712753 ,
102.68948 , 0.7165117, 1.157669 , 2.1900623, 2.189616 ,
1.0546725, 1.2338142, 0.5119512, 1.7162334, 1.6656369,
1.2711622, 1.2308022, 2.2867224], dtype=float32)
```
In the example shown, the first three lines import the required modules: module `spectrophore` for the calculation of spectrophores, module `Chem` to generate a RDKit molecule from a smiles string, and module `AllChem` to generate a 3D-conformation from the molecule. Next, a molecule is created from a smiles string (line 4), and a conformation is then generated at the 6th line after adding hydrogen atoms on the 5th line. Finally, on lines 7 and 8, a `SpectrophoreCalculator` object is generated and this object is then used to calculate a `spectrophore` descriptor (line 8), which consists in its default form of 4 * 12 numbers.
> Note: a few words on the shape of a `spectrophore`:
>
> Each `spectrophore` consists of a set of floating point numbers, and this set is always a multiple of 4 (the number of properties). The actual number count depends on how stereochemistry is treated in the calculation of spectrophores; this is controlled by the `stereo()` method:
- `stereo("none")`: the total number of numbers in a `spectrophore` is 48 (4 * 12; default),
- `stereo("unique")`: the total number of numbers in a `spectrophore` is 72 (4 * 18),
- `stereo("mirror")`: the total number of numbers in a `spectrophore` is 72 (4 * 18),
- `stereo("all")`: the total number of numbers in a `spectrophore` is 144 (4 * 36).
>
> Whatever the actual number of `spectrophore` points, these are always calculated according the same atomic properties. For example, consider a `spectrophore` of 4*n* points, then these points represent the following:
- **Points 1 to *n***: representing the interaction energies between the **atomic partial charges** and each of the *n* boxes;
- **Points *n*+1 to 2*n***: representing the interaction energies between the **atomic lipophilicities** and each of the *n* boxes;
- **Points 2*n*+1 to 3*n***: representing the interaction energies between the **atomic shape deviations** and each of the *n* boxes;
- **Points 3*n*+1 to 4*n***: representing the interaction energies between the **atomic electrophilicities** and each of the *n* boxes.
>
> Please have a look at the original publication form more information about the way these interaction energies are calculated, and what the `stereo()` method actually means.
If a molecule contains more than one 3D-conformation, then one may specify which conformation should be used for the calculation of `spectrophores`. As an example, consider the following code:
```python
>>> calculator = spectrophore.SpectrophoreCalculator(normalization='none')
Probes initialised: 48 number of probes in total
12 probes are used due to the imposed stereo flag
>>> aspirin = Chem.MolFromSmiles("CC(Oc1ccccc1C(O)=O)=O")
>>> aspirin = Chem.AddHs(aspirin)
>>> cids = AllChem.EmbedMultipleConfs(aspirin, numConfs=3, randomSeed=1)
>>> print(len(cids))
3
>>> for i in range(len(cids)): calculator.calculate(aspirin, i)
...
array([ 3.9325106, 5.057322 , 3.4856937, 5.479296 , 5.4948115,
6.778553 , 3.3464622, 4.14602 , 5.3836374, 7.561534 ,
6.8232827, 5.989737 , 7.6784873, 8.210138 , 7.167798 ,
10.159893 , 21.949173 , 21.82916 , 14.773877 , 17.42354 ,
19.190207 , 23.495762 , 27.787817 , 16.30314 , 216.26787 ,
333.40674 , 342.49576 , 360.4995 , 314.15094 , 339.17233 ,
154.0079 , 332.78748 , 327.1066 , 362.64258 , 318.9287 ,
409.1051 , 1.9632099, 3.081648 , 3.7438314, 4.0490594,
3.3057344, 3.7012224, 1.5342569, 3.5204902, 3.414244 ,
3.5912938, 3.358138 , 4.4090977], dtype=float32)
array([ 3.3760934, 6.2265506, 4.525822 , 4.819302 , 5.8221617,
5.351173 , 4.5672374, 5.7664223, 6.5604424, 8.637403 ,
8.085242 , 5.02033 , 6.9545755, 7.845039 , 7.3663807,
10.209895 , 22.197407 , 20.808294 , 14.168869 , 17.2765 ,
18.936844 , 23.069283 , 26.431952 , 16.029125 , 194.90318 ,
353.951 , 286.0087 , 303.20142 , 244.75073 , 268.89917 ,
169.84006 , 299.00415 , 282.5781 , 321.46698 , 344.0466 ,
359.3506 , 1.7995539, 3.2626312, 3.2555926, 3.4746647,
2.682269 , 2.511837 , 1.6443442, 3.2317195, 2.7628746,
3.0616467, 3.5081995, 3.877851 ], dtype=float32)
array([ 3.6380727, 5.0543313, 3.4793944, 5.1971545, 5.2814097,
6.6500306, 3.5975058, 3.8349445, 5.3234606, 7.27565 ,
6.456969 , 5.91822 , 7.750357 , 8.797789 , 7.37978 ,
10.291023 , 20.842691 , 21.349953 , 14.5347595, 17.440334 ,
19.158358 , 23.11196 , 27.614624 , 15.990323 , 204.78897 ,
325.7338 , 321.78525 , 338.5145 , 267.19482 , 374.50073 ,
142.66623 , 303.51685 , 296.4746 , 343.62027 , 324.18774 ,
394.01352 , 1.8762052, 3.0206196, 3.4741364, 3.7698758,
2.7347808, 3.719328 , 1.4108949, 3.237199 , 3.2545724,
3.386793 , 3.4655266, 4.238542 ], dtype=float32)
```
One can easily visualise `spectrophores` by plotting the actual values using `matplotlib`. For example, consider the following snippet:
```python
>>> import matplotlib.pyplot as plt
>>> mol = Chem.MolFromSmiles("CC(CCC1=CC=CC=C1Cl)N1CCOCC1")
>>> mol = Chem.AddHs(mol)
>>> cids = AllChem.EmbedMultipleConfs(mol, numConfs = 10, randomSeed = 1)
>>> spectrophores = []
>>> for cid in cids: spectrophores.append(calculator.calculate(mol, cid))
...
>>> for i in range(len(spectrophores)): plt.plot(range(1,49), spectrophores[i], label='Conf %d' % (i+1))
...
[<matplotlib.lines.Line2D object at 0x7f989dc3f4f0>]
[<matplotlib.lines.Line2D object at 0x7f989dc3f850>]
[<matplotlib.lines.Line2D object at 0x7f989dc3fbb0>]
[<matplotlib.lines.Line2D object at 0x7f989dc3ff10>]
[<matplotlib.lines.Line2D object at 0x7f989dc512b0>]
[<matplotlib.lines.Line2D object at 0x7f989dc51610>]
[<matplotlib.lines.Line2D object at 0x7f989dc51970>]
[<matplotlib.lines.Line2D object at 0x7f989dc51cd0>]
[<matplotlib.lines.Line2D object at 0x7f989dc5e070>]
[<matplotlib.lines.Line2D object at 0x7f989dc5e3d0>]
>>> plt.legend(loc='upper left')
<matplotlib.legend.Legend object at 0x7f9899470e80>
>>> plt.grid()
>>> plt.savefig("spectrophore/images/exampleplot1.png")
```
which generates the following plot:
Similarly, one can easily compare the `spectrophores` from two different molecules, and quantify the difference:
```python
>>> plt.close()
>>> spectrophores = []
>>> mols = [Chem.MolFromSmiles("ClC(Br)(I)F"), Chem.MolFromSmiles("CC(CCC1=CC=CC=C1Cl)N1CCOCC1")]
>>> for i in range(2):
... mols[i] = Chem.AddHs(mols[i])
... AllChem.EmbedMolecule(mols[i], randomSeed=1)
... spectrophores.append(calculator.calculate(mols[i]))
...
0
0
>>> for i in range(2): plt.plot(range(1,49), spectrophores[i], label='Molecule %d' % (i+1))
...
[<matplotlib.lines.Line2D object at 0x7faf420df520>]
[<matplotlib.lines.Line2D object at 0x7faf420df4c0>]
>>> plt.grid()
>>> plt.savefig("spectrophore/images/exampleplot2.png")
>>> from scipy.spatial import distance
>>> distance.euclidean(spectrophores[0],spectrophores[1])
2129.031982421875
```
From the last example, it is clear that the actual `spectrophore` values may differ a lot depending on the type of molecule. Also, the absolute values are depending on the property type, with some properties leading to large values (e.g. shape deviation) and others very small. For this reason, a number of normalization methods are provided as shown below.
### 2.2. Module methods
#### resolution
The `resolution()` method controls the smallest distance between the molecule and the surrounding box. By default this value is set to 3.0 A. The `resolution()` can be specified at the moment of class creation, or later on using the `resolution()` method:
```python
>>> mol = Chem.MolFromSmiles("ClC(Br)(I)F")
>>> AllChem.EmbedMolecule(mol, randomSeed=1)
0
>>> calculator = spectrophore.SpectrophoreCalculator(normalization='none') # Default of 3.0
Probes initialised: 48 number of probes in total
12 probes are used due to the imposed stereo flag
>>> print(calculator.calculate(mol)[0])
2.9870005
>>> calculator = spectrophore.SpectrophoreCalculator(normalization='none', resolution = 5.0)
Probes initialised: 48 number of probes in total
12 probes are used due to the imposed stereo flag
>>> print(calculator.calculate(mol)[0])
1.3418818
>>> calculator.resolution(10.0)
>>> print(calculator.calculate(mol)[0])
0.33471847
```
The larger the resolution value (e.g. 10.0 *versus* 3.0 A), the smaller the interaction energies and corresponding `spectrophore` values.
Calling the `resolution()` method without an argument returns the current resolution value:
```python
>>> calculator.resolution()
10.0
```
#### accuracy
The `accuracy()` method controls the angular stepsize by which the molecule is rotated within the cages. By default this value is set to 20°. This parameter can be modified either at class creation, or using the `accuracy()` method later on. The accuracy should be an integer fraction of 180, hence 180 modulus *accuracy* should be equal to 0. The smaller the accuracy value (meaning smaller angular stepsizes), the longer the computation time:
```python
>>> calculator = spectrophore.SpectrophoreCalculator(accuracy = 20.0, normalization = 'none') # Default
Probes initialised: 48 number of probes in total
12 probes are used due to the imposed stereo flag
>>> print(calculator.calculate(mol)[0])
2.9870005
>>> calculator = spectrophore.SpectrophoreCalculator(accuracy = 2.0, normalization = 'none')
Probes initialised: 48 number of probes in total
Only using 12 probes
>>> print(calculator.calculate(mol)[0]) # Takes some time
3.0315506
>>> 100.0 * (3.0315506 - 2.9870005) / 3.0315506
1.469548289908139
```
Calling the `accuracy()` method without an argument returns the current accuracy value:
```python
>>> calculator.accuracy()
np.int64(2)
```
#### normalization
With the `normalization()` method, one can specify the type of `spectrophore` normalization. There are four possibilities:
- `normalization("none")`: no normalization is applied and the `spectrophore` values are the raw calculated interaction energies (multiplied by -100),
- `normalization("mean")`: for each property, the average value is calculated and each of the individual `spectrophore` property value are reduced by these mean values. This centers the calculated values around 0,
- `normalization("std")`: for each property, the standard deviation is calculated and each of the individual `spectrophore` property value is divided by these standard deviations, making the standard deviation 1
- `normalization("all")`: each spectrophore value is normalized by mean and standard deviation. This is the default option.
The default value is "all".
```python
>>> calculator.accuracy(20)
>>> calculator.normalization("none")
>>> spec = calculator.calculate(mol)
>>> print(spec[:12])
[2.9870005 2.7023234 1.8029699 4.468908 7.375546 7.252276 4.1123323
4.0559926 5.908458 8.5649605 9.328256 6.2296877]
>>> sum(spec[:12])
np.float32(64.78871)
>>> calculator.normalization("mean")
>>> spec = calculator.calculate(mol)
>>> print(spec[:12])
[-2.4120588 -2.6967359 -3.5960894 -0.93015146 1.9764867 1.8532166
-1.286727 -1.3430667 0.50939894 3.1659012 3.9291964 0.8306284 ]
>>> sum(spec[:12])
np.float32(-4.7683716e-07)
>>> calculator.normalization("std")
>>> spec = calculator.calculate(mol)
>>> print(spec[:12])
[1.2924119 1.1692382 0.78010696 1.933602 3.1912427 3.1379066
1.7793193 1.7549422 2.556465 3.7058773 4.036139 2.6954541 ]
>>> sum(spec[:12])
np.float32(28.032705)
>>> calculator.normalization("all")
>>> spec = calculator.calculate(mol)
>>> print(spec[:12])
[-1.0436468 -1.1668205 -1.5559518 -0.40245688 0.855184 0.80184764
-0.55673957 -0.58111656 0.22040614 1.3698184 1.7000802 0.35939535]
>>> sum(spec[:12])
np.float32(-6.2584877e-07)
```
Using a normalization over 'all' makes it more easier to compare `spectrophores` between molecules:
```python
>>> plt.close()
>>> mols = [Chem.MolFromSmiles("ClC(Br)(I)F"), Chem.MolFromSmiles("CC(CCC1=CC=CC=C1Cl)N1CCOCC1")]
>>> spectrophores = []
>>> for i in range(2):
... mols[i] = Chem.AddHs(mols[i])
... AllChem.EmbedMolecule(mols[i], randomSeed = 1)
... spectrophores.append(calculator.calculate(mols[i]))
...
0
0
>>> for i in range(2): plt.plot(range(1,49), spectrophores[i], label='Molecule %d' % (i+1))
...
[<matplotlib.lines.Line2D object at 0x744aeb797770>]
[<matplotlib.lines.Line2D object at 0x744aeb797aa0>]
>>> plt.legend()
<matplotlib.legend.Legend object at 0x744af8519640>
>>> plt.grid()
>>> plt.savefig("spectrophore/images/exampleplot3.png")
>>> from scipy.spatial import distance
>>> distance.euclidean(spectrophores[0],spectrophores[1])
8.419559478759766
```
The same holds true when comparing `spectrophores` from different conformations:
```python
>>> plt.close()
>>> spectrophores = []
>>> mol = Chem.MolFromSmiles("CC(CCC1=CC=CC=C1Cl)N1CCOCC1")
>>> mol = Chem.AddHs(mol)
>>> cids = AllChem.EmbedMultipleConfs(mol, numConfs = 10, randomSeed = 1)
>>> calculator.normalization("all")
>>> for cid in cids: spectrophores.append(calculator.calculate(mol, cid))
...
>>> for i in range(len(spectrophores)): plt.plot(range(1,49), spectrophores[i], label='Conf %d' % (i+1))
...
[<matplotlib.lines.Line2D object at 0x744aeb7ee000>]
[<matplotlib.lines.Line2D object at 0x744aeb7c6780>]
[<matplotlib.lines.Line2D object at 0x744aeb7c43e0>]
[<matplotlib.lines.Line2D object at 0x744aeb7c5eb0>]
[<matplotlib.lines.Line2D object at 0x744aeb7c51f0>]
[<matplotlib.lines.Line2D object at 0x744aeb7c5340>]
[<matplotlib.lines.Line2D object at 0x744aeb7c5a30>]
[<matplotlib.lines.Line2D object at 0x744aeb7c6180>]
[<matplotlib.lines.Line2D object at 0x744aeb7c5e20>]
[<matplotlib.lines.Line2D object at 0x744aeb7c70e0>]
>>> plt.legend(loc='upper left')
<matplotlib.legend.Legend object at 0x744aeb8291c0>
>>> plt.savefig("spectrophore/images/exampleplot4.png")
>>> from scipy.spatial import distance
>>> distance.euclidean(spectrophores[0],spectrophores[1])
6.242583751678467
>>> distance.euclidean(spectrophores[0],spectrophores[2])
6.020205497741699
>>> distance.euclidean(spectrophores[1],spectrophores[2])
4.005795001983643
```
#### stereo
The `stereo()` method specifies the kind of cages to be used. The reason for this is that some of the cages that are used to calculate `spectrophores` have a stereospecific distribution of the interaction points:
There are four possibilities:
- `stereo("none")`: no stereospecificity (default). `Spectrophores` are generated using cages that are not stereospecific. For most applications, these `spectrophores` will suffice
- `stereo("unique")`: unique stereospecificity. `Spectrophores` are generated using unique stereospecific cages
- `stereo("mirror")`: mirror stereospecificity. Mirror stereospecific `spectrophores` are `spectrophores` resulting from the mirror enantiomeric form of the input molecules
- `stereo("all")`: all cages are used. This results in the longest `spectrophores` and should only in specific cases be used
The differences between the corresponding data points of unique and mirror stereospecific `spectrophores` are very small and require very long calculation times to obtain a sufficiently high quality level. This increased quality level is triggered by the `accuracy` setting and will result in calculation times being increased by at least a factor 100. As a consequence, it is recommended to apply this increased accuracy only in combination with a limited number of molecules, and when the small differences between the stereospecific `spectrophores` are really critical. However, for the vast majority of virtual screening applications, this increased accuracy is not required as long as it is not the intention to draw conclusions about differences in the underlying molecular stereoselectivity. Non-stereospecific `spectrophores` will therefore suffice for most applications.
#### mode
The `mode()` method currently has two supported options:
- `mode("classic")`: This is the default option, and computes spectrophores as described in the original [manuscript](https://jcheminf.biomedcentral.com/articles/10.1186/s13321-018-0268-9).
- `mode("full")`: This option keeps all the data from every frame, which could potentially be useful in down-stream applications. The output is a 2D NumPy array, where:
- The first dimension corresponds to the number of frames, which is determined by the accuracy option.
- The second dimension represents the length of the spectrophore, which is influenced by the stereo option.
Computing the column-wise minimum of a full spectrophore yields the corresponding classic spectrophore
## 3. Interpreting `spectrophores`
A `spectrophore` is a vector of real number and has a certain length. The length depends on the used `stereo` method and the number of properties. The standard setting uses a set of non-stereospecific probes in combination with four properties:
- property 1: atomic partial charges
- property 2: atomic lipophilicities
- property 3: atomic shape deviations
- property 4: atomic electrophilicties
The combination of four properties and the set of non-stereospecific probes leads to a `spectrophore` vector length of 48. The use of other probes leads to other vector lengths, as summarised in this table:
| Stereospecificity | Number of probes | Number of properties | Length |
| ----------------- |:----------------:|:--------------------:|:------:|
| none | 12 | 4 | 48 |
| unique | 18 | 4 | 72 |
| mirror | 18 | 4 | 72 |
| all | 36 | 4 | 144 |
The general layout of a `spectrophore`, irrespective of its length, is always:
| Property 1 | Property 2 | Property 3 | Property 4 |
|:------------------:|:------------------:|:------------------:|:------------------:|
| probe 1..probe *n* | probe 1..probe *n* | probe 1..probe *n* | probe 1..probe *n* |
meaning that the first *n* values (with *n* being the number of probes) are calculated using property 1 (partial charges), then another *n* values (*n*+1 up to 2*n*) calculated using property 2 (lipophilicities), and so forth.
## 4. Release updates
- 1.3.0:
- Added the mode() method to the SpectrophoreCalculator class, with 'classic' and 'full' as options
- Added the calculate_string() method to the SpectrophoreCalculator class
- Added CI using GitHub Actions
- Updated setup.py for easier install using pip
- Moved the main function to a seperate main.py, making the standalone functionality more easily accessible using the spectrophore alias
- Moved the Jupyter notebook to the main repo and updated it to the new code
- Updated readme
- General code cleanup
- 1.2.0:
- Switched to numpy.float32 type to achieve major speedup
- 1.1.0:
- Updated and optimised the NumPy code
- Bug fixes
- Introduced Numba
- Made the 'all' normalization method the default one
- Added a test suite
- 1.0.1: First official release on PyPi
## 5. Reference and citation
If you use the `spectrophore` technology in your own research work, please cite as follows:
Gladysz, R.; Mendes Dos Santos, F.; Langenaeker, W.; Thijs, G.; Augustyns, K.; De Winter, H. (2018) 'Spectrophores as one-dimensional descriptors calculated from three-dimensional atomic properties: applications ranging from scaffold hopping to multi-target virtual screening', *J. Cheminformatics* **10**, 9.
Raw data
{
"_id": null,
"home_page": "https://github.com/UAMCAntwerpen/spectrophore",
"name": "uamc-spectrophore",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.9",
"maintainer_email": null,
"keywords": "uamc, spectrophore, rdkit, cheminformatics",
"author": "Hans De Winter",
"author_email": "hans.dewinter@uantwerpen.be",
"download_url": "https://files.pythonhosted.org/packages/73/63/c87d71194c934214ec0c14de3725082ea9d0b3fad38a2583750fa746ab88/uamc_spectrophore-1.3.0.tar.gz",
"platform": null,
"description": "# Spectrophores\n[](https://doi.org/10.1186/s13321-018-0268-9) [](https://github.com/silicos-it/spectrophore/actions/workflows/test.yml) \n\nThis module contains python code to calculate `spectrophores` from molecules. It is uses `RDKit` to process molecules, and `numba` to improve the performance.\n\nThe technology and its applications have been described in [*Journal of Cheminformatics* (2018) **10**, 9](https://jcheminf.biomedcentral.com/articles/10.1186/s13321-018-0268-9). The paper is also included in this distribution.\n\n\nThe `spectrophore` code can be used in two ways:\n- As a standalone program to convert the molecules in a sd-file into their corresponding `spectrophores`\n- As a `python` module to import in your own `python` code.\n\nIn the following sections, both usages will be documented.\n\nThis repository also includes `spectrophore_notebook.ipynb`, a portable Jupyter Notebook which could be useful for quick and interactive data analysis with spectrophores.\n\n## Installation\n\n### Installation using pip (recommended)\n1. Create and activate a new Anaconda environment\n \n ```\n conda create --name spectrophore python=3\n ```\n ```\n conda activate spectrophore\n ```\n2. Install the repo via pip. The RDKit, Numba and other dependencies should be installed automatically.\n \n ```\n pip install uamc-spectrophore\n ```\n### Installation from Source\n1. Clone the repository and navigate to it\n ```\n git clone https://github.com/silicos-it/spectrophore.git\n ```\n ```\n cd spectrophore\n ```\n2. Create and activate a new Anaconda environment\n ```\n conda create --name spectrophore python=3\n ```\n ```\n conda activate spectrophore\n ```\n3. Install the repo by using pip. The RDKit, Numba and other dependencies should be installed automatically.\n ```\n pip install .\n ```\n #### For Contributors\n\n To install with additional development dependencies (`pytest`, `ruff`, etc.), use:\n ```\n pip install .[develop]\n ```\n\n## Usage\n### 1. Usage as a standalone program\n\nAfter installation, one should be able to use the `spectrophore` package as a standalone program to calculate `spectrophores` from a sd-file with molecules.\n\n```console\n> python -m spectrophore -h\n```\nor even more convenient:\n\n```console\n> spectrophore -h\n```\n\nThis will provide you with all details on how to calculate `spectrophores` from a sd-file, as well as all the configurable settings:\n\n```console\nusage: spectrophore [-h] [-n {none,mean,all,std}] [-m {classic,full}] [-s {none,unique,mirror,all}] [-a {1,2,3,4,5,6,9,10,12,15,18,20,30,36,45,60,90,180}] [-r RESOLUTION] [-p MAX_WORKERS]\n [--silent] -i INFILE -o OUTFILE\n\nConvert a SDF-file containing 3D structures to a csv with the corresponding spectrophores fingerprints\n\noptions:\n -h, --help show this help message and exit\n -n {none,mean,all,std}, --norm {none,mean,all,std}\n Normalization setting (default: all)\n -m {classic,full}, --mode {classic,full}\n Which type of spectrophore to calculate (default: classic)\n -s {none,unique,mirror,all}, --stereo {none,unique,mirror,all}\n Stereo setting (default: none)\n -a {1,2,3,4,5,6,9,10,12,15,18,20,30,36,45,60,90,180}, --accuracy {1,2,3,4,5,6,9,10,12,15,18,20,30,36,45,60,90,180}\n Accuracy setting (default: 20)\n -r RESOLUTION, --resolution RESOLUTION\n Resolution setting (>0) (default: 3)\n -p MAX_WORKERS, --np MAX_WORKERS\n Number of processors to use; -1 is all processors (default: -1)\n --silent Don't show a progressbar (default: False)\n\nrequired arguments:\n -i INFILE, --in INFILE\n Input sdf file (default: None)\n -o OUTFILE, --out OUTFILE\n Output spectrophore file (default: None)\n```\n\n\n### 2. Usage as a python module\n\n### 2.1. Introduction\n\nOnce you have installed the `uamc-spectrophore` package, you are ready to calculate `spectrophores`. In its most simple form, `spectrophores` can be computed as follows:\n\n```python\n>>> from spectrophore import spectrophore\n>>> from rdkit import Chem\n>>> from rdkit.Chem import AllChem\n>>> mol = Chem.MolFromSmiles(\"c1ncncc1\")\n>>> mol = Chem.AddHs(mol)\n>>> AllChem.EmbedMolecule(mol, randomSeed=1)\n0\n>>> calculator = spectrophore.SpectrophoreCalculator(normalization='none')\nProbes initialised: 48 number of probes in total\n12 probes are used due to the imposed stereo flag\n>>> calculator.calculate(mol)\narray([ 1.4826062, 2.0534244, 1.6182035, 3.1277552, 2.5039089,\n 5.157424 , 2.3022916, 1.6652936, 3.4908977, 4.0566797,\n 5.020808 , 3.0757458, 0.6589167, 0.7786141, 4.324721 ,\n 4.3273506, 2.928256 , 4.132863 , 1.9475778, 3.5233636,\n 4.6044445, 4.1554613, 4.5122004, 5.2647495, 46.877754 ,\n 71.05832 , 96.21371 , 94.29615 , 51.998947 , 63.523525 ,\n 24.393661 , 80.728264 , 73.68167 , 65.27555 , 56.712753 ,\n 102.68948 , 0.7165117, 1.157669 , 2.1900623, 2.189616 ,\n 1.0546725, 1.2338142, 0.5119512, 1.7162334, 1.6656369,\n 1.2711622, 1.2308022, 2.2867224], dtype=float32)\n```\n\nIn the example shown, the first three lines import the required modules: module `spectrophore` for the calculation of spectrophores, module `Chem` to generate a RDKit molecule from a smiles string, and module `AllChem` to generate a 3D-conformation from the molecule. Next, a molecule is created from a smiles string (line 4), and a conformation is then generated at the 6th line after adding hydrogen atoms on the 5th line. Finally, on lines 7 and 8, a `SpectrophoreCalculator` object is generated and this object is then used to calculate a `spectrophore` descriptor (line 8), which consists in its default form of 4 * 12 numbers.\n\n> Note: a few words on the shape of a `spectrophore`:\n>\n> Each `spectrophore` consists of a set of floating point numbers, and this set is always a multiple of 4 (the number of properties). The actual number count depends on how stereochemistry is treated in the calculation of spectrophores; this is controlled by the `stereo()` method:\n- `stereo(\"none\")`: the total number of numbers in a `spectrophore` is 48 (4 * 12; default),\n- `stereo(\"unique\")`: the total number of numbers in a `spectrophore` is 72 (4 * 18),\n- `stereo(\"mirror\")`: the total number of numbers in a `spectrophore` is 72 (4 * 18),\n- `stereo(\"all\")`: the total number of numbers in a `spectrophore` is 144 (4 * 36).\n>\n> Whatever the actual number of `spectrophore` points, these are always calculated according the same atomic properties. For example, consider a `spectrophore` of 4*n* points, then these points represent the following:\n- **Points 1 to *n***: representing the interaction energies between the **atomic partial charges** and each of the *n* boxes;\n- **Points *n*+1 to 2*n***: representing the interaction energies between the **atomic lipophilicities** and each of the *n* boxes;\n- **Points 2*n*+1 to 3*n***: representing the interaction energies between the **atomic shape deviations** and each of the *n* boxes;\n- **Points 3*n*+1 to 4*n***: representing the interaction energies between the **atomic electrophilicities** and each of the *n* boxes.\n>\n> Please have a look at the original publication form more information about the way these interaction energies are calculated, and what the `stereo()` method actually means.\n\nIf a molecule contains more than one 3D-conformation, then one may specify which conformation should be used for the calculation of `spectrophores`. As an example, consider the following code:\n\n```python\n>>> calculator = spectrophore.SpectrophoreCalculator(normalization='none')\nProbes initialised: 48 number of probes in total\n12 probes are used due to the imposed stereo flag\n>>> aspirin = Chem.MolFromSmiles(\"CC(Oc1ccccc1C(O)=O)=O\")\n>>> aspirin = Chem.AddHs(aspirin)\n>>> cids = AllChem.EmbedMultipleConfs(aspirin, numConfs=3, randomSeed=1)\n>>> print(len(cids))\n3\n>>> for i in range(len(cids)): calculator.calculate(aspirin, i)\n...\narray([ 3.9325106, 5.057322 , 3.4856937, 5.479296 , 5.4948115,\n 6.778553 , 3.3464622, 4.14602 , 5.3836374, 7.561534 ,\n 6.8232827, 5.989737 , 7.6784873, 8.210138 , 7.167798 ,\n 10.159893 , 21.949173 , 21.82916 , 14.773877 , 17.42354 ,\n 19.190207 , 23.495762 , 27.787817 , 16.30314 , 216.26787 ,\n 333.40674 , 342.49576 , 360.4995 , 314.15094 , 339.17233 ,\n 154.0079 , 332.78748 , 327.1066 , 362.64258 , 318.9287 ,\n 409.1051 , 1.9632099, 3.081648 , 3.7438314, 4.0490594,\n 3.3057344, 3.7012224, 1.5342569, 3.5204902, 3.414244 ,\n 3.5912938, 3.358138 , 4.4090977], dtype=float32)\narray([ 3.3760934, 6.2265506, 4.525822 , 4.819302 , 5.8221617,\n 5.351173 , 4.5672374, 5.7664223, 6.5604424, 8.637403 ,\n 8.085242 , 5.02033 , 6.9545755, 7.845039 , 7.3663807,\n 10.209895 , 22.197407 , 20.808294 , 14.168869 , 17.2765 ,\n 18.936844 , 23.069283 , 26.431952 , 16.029125 , 194.90318 ,\n 353.951 , 286.0087 , 303.20142 , 244.75073 , 268.89917 ,\n 169.84006 , 299.00415 , 282.5781 , 321.46698 , 344.0466 ,\n 359.3506 , 1.7995539, 3.2626312, 3.2555926, 3.4746647,\n 2.682269 , 2.511837 , 1.6443442, 3.2317195, 2.7628746,\n 3.0616467, 3.5081995, 3.877851 ], dtype=float32)\narray([ 3.6380727, 5.0543313, 3.4793944, 5.1971545, 5.2814097,\n 6.6500306, 3.5975058, 3.8349445, 5.3234606, 7.27565 ,\n 6.456969 , 5.91822 , 7.750357 , 8.797789 , 7.37978 ,\n 10.291023 , 20.842691 , 21.349953 , 14.5347595, 17.440334 ,\n 19.158358 , 23.11196 , 27.614624 , 15.990323 , 204.78897 ,\n 325.7338 , 321.78525 , 338.5145 , 267.19482 , 374.50073 ,\n 142.66623 , 303.51685 , 296.4746 , 343.62027 , 324.18774 ,\n 394.01352 , 1.8762052, 3.0206196, 3.4741364, 3.7698758,\n 2.7347808, 3.719328 , 1.4108949, 3.237199 , 3.2545724,\n 3.386793 , 3.4655266, 4.238542 ], dtype=float32)\n```\n\nOne can easily visualise `spectrophores` by plotting the actual values using `matplotlib`. For example, consider the following snippet:\n\n```python\n>>> import matplotlib.pyplot as plt\n>>> mol = Chem.MolFromSmiles(\"CC(CCC1=CC=CC=C1Cl)N1CCOCC1\")\n>>> mol = Chem.AddHs(mol)\n>>> cids = AllChem.EmbedMultipleConfs(mol, numConfs = 10, randomSeed = 1)\n>>> spectrophores = []\n>>> for cid in cids: spectrophores.append(calculator.calculate(mol, cid))\n...\n>>> for i in range(len(spectrophores)): plt.plot(range(1,49), spectrophores[i], label='Conf %d' % (i+1))\n...\n[<matplotlib.lines.Line2D object at 0x7f989dc3f4f0>]\n[<matplotlib.lines.Line2D object at 0x7f989dc3f850>]\n[<matplotlib.lines.Line2D object at 0x7f989dc3fbb0>]\n[<matplotlib.lines.Line2D object at 0x7f989dc3ff10>]\n[<matplotlib.lines.Line2D object at 0x7f989dc512b0>]\n[<matplotlib.lines.Line2D object at 0x7f989dc51610>]\n[<matplotlib.lines.Line2D object at 0x7f989dc51970>]\n[<matplotlib.lines.Line2D object at 0x7f989dc51cd0>]\n[<matplotlib.lines.Line2D object at 0x7f989dc5e070>]\n[<matplotlib.lines.Line2D object at 0x7f989dc5e3d0>]\n>>> plt.legend(loc='upper left')\n<matplotlib.legend.Legend object at 0x7f9899470e80>\n>>> plt.grid()\n>>> plt.savefig(\"spectrophore/images/exampleplot1.png\")\n```\n\nwhich generates the following plot:\n\n\n\nSimilarly, one can easily compare the `spectrophores` from two different molecules, and quantify the difference:\n\n```python\n>>> plt.close()\n>>> spectrophores = []\n>>> mols = [Chem.MolFromSmiles(\"ClC(Br)(I)F\"), Chem.MolFromSmiles(\"CC(CCC1=CC=CC=C1Cl)N1CCOCC1\")]\n>>> for i in range(2):\n... mols[i] = Chem.AddHs(mols[i])\n... AllChem.EmbedMolecule(mols[i], randomSeed=1)\n... spectrophores.append(calculator.calculate(mols[i]))\n...\n0\n0\n>>> for i in range(2): plt.plot(range(1,49), spectrophores[i], label='Molecule %d' % (i+1))\n...\n[<matplotlib.lines.Line2D object at 0x7faf420df520>]\n[<matplotlib.lines.Line2D object at 0x7faf420df4c0>]\n>>> plt.grid()\n>>> plt.savefig(\"spectrophore/images/exampleplot2.png\")\n>>> from scipy.spatial import distance\n>>> distance.euclidean(spectrophores[0],spectrophores[1])\n2129.031982421875\n```\n\n\n\n\nFrom the last example, it is clear that the actual `spectrophore` values may differ a lot depending on the type of molecule. Also, the absolute values are depending on the property type, with some properties leading to large values (e.g. shape deviation) and others very small. For this reason, a number of normalization methods are provided as shown below.\n\n\n### 2.2. Module methods\n\n#### resolution\n\nThe `resolution()` method controls the smallest distance between the molecule and the surrounding box. By default this value is set to 3.0 A. The `resolution()` can be specified at the moment of class creation, or later on using the `resolution()` method:\n\n```python\n>>> mol = Chem.MolFromSmiles(\"ClC(Br)(I)F\")\n>>> AllChem.EmbedMolecule(mol, randomSeed=1)\n0\n>>> calculator = spectrophore.SpectrophoreCalculator(normalization='none') # Default of 3.0\nProbes initialised: 48 number of probes in total\n12 probes are used due to the imposed stereo flag\n>>> print(calculator.calculate(mol)[0])\n2.9870005\n>>> calculator = spectrophore.SpectrophoreCalculator(normalization='none', resolution = 5.0)\nProbes initialised: 48 number of probes in total\n12 probes are used due to the imposed stereo flag\n>>> print(calculator.calculate(mol)[0])\n1.3418818\n>>> calculator.resolution(10.0)\n>>> print(calculator.calculate(mol)[0])\n0.33471847\n```\n\nThe larger the resolution value (e.g. 10.0 *versus* 3.0 A), the smaller the interaction energies and corresponding `spectrophore` values.\n\nCalling the `resolution()` method without an argument returns the current resolution value:\n\n```python\n>>> calculator.resolution()\n10.0\n```\n\n\n#### accuracy\n\nThe `accuracy()` method controls the angular stepsize by which the molecule is rotated within the cages. By default this value is set to 20\u00c2\u00b0. This parameter can be modified either at class creation, or using the `accuracy()` method later on. The accuracy should be an integer fraction of 180, hence 180 modulus *accuracy* should be equal to 0. The smaller the accuracy value (meaning smaller angular stepsizes), the longer the computation time:\n\n```python\n>>> calculator = spectrophore.SpectrophoreCalculator(accuracy = 20.0, normalization = 'none') # Default\nProbes initialised: 48 number of probes in total\n12 probes are used due to the imposed stereo flag\n>>> print(calculator.calculate(mol)[0])\n2.9870005\n>>> calculator = spectrophore.SpectrophoreCalculator(accuracy = 2.0, normalization = 'none')\nProbes initialised: 48 number of probes in total\nOnly using 12 probes\n>>> print(calculator.calculate(mol)[0]) # Takes some time\n3.0315506\n>>> 100.0 * (3.0315506 - 2.9870005) / 3.0315506\n1.469548289908139\n```\n\nCalling the `accuracy()` method without an argument returns the current accuracy value:\n\n```python\n>>> calculator.accuracy()\nnp.int64(2)\n```\n\n#### normalization\n\nWith the `normalization()` method, one can specify the type of `spectrophore` normalization. There are four possibilities:\n- `normalization(\"none\")`: no normalization is applied and the `spectrophore` values are the raw calculated interaction energies (multiplied by -100),\n- `normalization(\"mean\")`: for each property, the average value is calculated and each of the individual `spectrophore` property value are reduced by these mean values. This centers the calculated values around 0,\n- `normalization(\"std\")`: for each property, the standard deviation is calculated and each of the individual `spectrophore` property value is divided by these standard deviations, making the standard deviation 1\n- `normalization(\"all\")`: each spectrophore value is normalized by mean and standard deviation. This is the default option.\n\nThe default value is \"all\".\n\n```python\n>>> calculator.accuracy(20)\n>>> calculator.normalization(\"none\")\n>>> spec = calculator.calculate(mol)\n>>> print(spec[:12])\n[2.9870005 2.7023234 1.8029699 4.468908 7.375546 7.252276 4.1123323\n 4.0559926 5.908458 8.5649605 9.328256 6.2296877]\n>>> sum(spec[:12])\nnp.float32(64.78871)\n>>> calculator.normalization(\"mean\")\n>>> spec = calculator.calculate(mol)\n>>> print(spec[:12])\n[-2.4120588 -2.6967359 -3.5960894 -0.93015146 1.9764867 1.8532166\n -1.286727 -1.3430667 0.50939894 3.1659012 3.9291964 0.8306284 ]\n>>> sum(spec[:12])\nnp.float32(-4.7683716e-07)\n>>> calculator.normalization(\"std\")\n>>> spec = calculator.calculate(mol)\n>>> print(spec[:12])\n[1.2924119 1.1692382 0.78010696 1.933602 3.1912427 3.1379066\n 1.7793193 1.7549422 2.556465 3.7058773 4.036139 2.6954541 ]\n>>> sum(spec[:12])\nnp.float32(28.032705)\n>>> calculator.normalization(\"all\")\n>>> spec = calculator.calculate(mol)\n>>> print(spec[:12])\n[-1.0436468 -1.1668205 -1.5559518 -0.40245688 0.855184 0.80184764\n -0.55673957 -0.58111656 0.22040614 1.3698184 1.7000802 0.35939535]\n>>> sum(spec[:12])\nnp.float32(-6.2584877e-07)\n```\n\nUsing a normalization over 'all' makes it more easier to compare `spectrophores` between molecules:\n\n```python\n>>> plt.close()\n>>> mols = [Chem.MolFromSmiles(\"ClC(Br)(I)F\"), Chem.MolFromSmiles(\"CC(CCC1=CC=CC=C1Cl)N1CCOCC1\")]\n>>> spectrophores = []\n>>> for i in range(2):\n... mols[i] = Chem.AddHs(mols[i])\n... AllChem.EmbedMolecule(mols[i], randomSeed = 1)\n... spectrophores.append(calculator.calculate(mols[i]))\n...\n0\n0\n>>> for i in range(2): plt.plot(range(1,49), spectrophores[i], label='Molecule %d' % (i+1))\n...\n[<matplotlib.lines.Line2D object at 0x744aeb797770>]\n[<matplotlib.lines.Line2D object at 0x744aeb797aa0>]\n>>> plt.legend()\n<matplotlib.legend.Legend object at 0x744af8519640>\n>>> plt.grid()\n>>> plt.savefig(\"spectrophore/images/exampleplot3.png\")\n>>> from scipy.spatial import distance\n>>> distance.euclidean(spectrophores[0],spectrophores[1])\n8.419559478759766\n```\n\n\n\nThe same holds true when comparing `spectrophores` from different conformations:\n\n```python\n>>> plt.close()\n>>> spectrophores = []\n>>> mol = Chem.MolFromSmiles(\"CC(CCC1=CC=CC=C1Cl)N1CCOCC1\")\n>>> mol = Chem.AddHs(mol)\n>>> cids = AllChem.EmbedMultipleConfs(mol, numConfs = 10, randomSeed = 1)\n>>> calculator.normalization(\"all\")\n>>> for cid in cids: spectrophores.append(calculator.calculate(mol, cid))\n...\n>>> for i in range(len(spectrophores)): plt.plot(range(1,49), spectrophores[i], label='Conf %d' % (i+1))\n...\n[<matplotlib.lines.Line2D object at 0x744aeb7ee000>]\n[<matplotlib.lines.Line2D object at 0x744aeb7c6780>]\n[<matplotlib.lines.Line2D object at 0x744aeb7c43e0>]\n[<matplotlib.lines.Line2D object at 0x744aeb7c5eb0>]\n[<matplotlib.lines.Line2D object at 0x744aeb7c51f0>]\n[<matplotlib.lines.Line2D object at 0x744aeb7c5340>]\n[<matplotlib.lines.Line2D object at 0x744aeb7c5a30>]\n[<matplotlib.lines.Line2D object at 0x744aeb7c6180>]\n[<matplotlib.lines.Line2D object at 0x744aeb7c5e20>]\n[<matplotlib.lines.Line2D object at 0x744aeb7c70e0>]\n\n>>> plt.legend(loc='upper left')\n<matplotlib.legend.Legend object at 0x744aeb8291c0>\n>>> plt.savefig(\"spectrophore/images/exampleplot4.png\")\n>>> from scipy.spatial import distance\n>>> distance.euclidean(spectrophores[0],spectrophores[1])\n6.242583751678467\n>>> distance.euclidean(spectrophores[0],spectrophores[2])\n6.020205497741699\n>>> distance.euclidean(spectrophores[1],spectrophores[2])\n4.005795001983643\n```\n\n\n\n\n\n#### stereo\n\nThe `stereo()` method specifies the kind of cages to be used. The reason for this is that some of the cages that are used to calculate `spectrophores` have a stereospecific distribution of the interaction points:\n\n\n\n\nThere are four possibilities:\n- `stereo(\"none\")`: no stereospecificity (default). `Spectrophores` are generated using cages that are not stereospecific. For most applications, these `spectrophores` will suffice\n- `stereo(\"unique\")`: unique stereospecificity. `Spectrophores` are generated using unique stereospecific cages\n- `stereo(\"mirror\")`: mirror stereospecificity. Mirror stereospecific `spectrophores` are `spectrophores` resulting from the mirror enantiomeric form of the input molecules\n- `stereo(\"all\")`: all cages are used. This results in the longest `spectrophores` and should only in specific cases be used\n\nThe differences between the corresponding data points of unique and mirror stereospecific `spectrophores` are very small and require very long calculation times to obtain a sufficiently high quality level. This increased quality level is triggered by the `accuracy` setting and will result in calculation times being increased by at least a factor 100. As a consequence, it is recommended to apply this increased accuracy only in combination with a limited number of molecules, and when the small differences between the stereospecific `spectrophores` are really critical. However, for the vast majority of virtual screening applications, this increased accuracy is not required as long as it is not the intention to draw conclusions about differences in the underlying molecular stereoselectivity. Non-stereospecific `spectrophores` will therefore suffice for most applications.\n\n#### mode\nThe `mode()` method currently has two supported options:\n- `mode(\"classic\")`: This is the default option, and computes spectrophores as described in the original [manuscript](https://jcheminf.biomedcentral.com/articles/10.1186/s13321-018-0268-9).\n- `mode(\"full\")`: This option keeps all the data from every frame, which could potentially be useful in down-stream applications. The output is a 2D NumPy array, where:\n - The first dimension corresponds to the number of frames, which is determined by the accuracy option.\n - The second dimension represents the length of the spectrophore, which is influenced by the stereo option.\n \nComputing the column-wise minimum of a full spectrophore yields the corresponding classic spectrophore \n\n## 3. Interpreting `spectrophores`\n\nA `spectrophore` is a vector of real number and has a certain length. The length depends on the used `stereo` method and the number of properties. The standard setting uses a set of non-stereospecific probes in combination with four properties:\n- property 1: atomic partial charges\n- property 2: atomic lipophilicities\n- property 3: atomic shape deviations\n- property 4: atomic electrophilicties\n\nThe combination of four properties and the set of non-stereospecific probes leads to a `spectrophore` vector length of 48. The use of other probes leads to other vector lengths, as summarised in this table:\n\n| Stereospecificity | Number of probes | Number of properties | Length |\n| ----------------- |:----------------:|:--------------------:|:------:|\n| none | 12 | 4 | 48 |\n| unique | 18 | 4 | 72 |\n| mirror | 18 | 4 | 72 |\n| all | 36 | 4 | 144 |\n\nThe general layout of a `spectrophore`, irrespective of its length, is always:\n\n| Property 1 | Property 2 | Property 3 | Property 4 |\n|:------------------:|:------------------:|:------------------:|:------------------:|\n| probe 1..probe *n* | probe 1..probe *n* | probe 1..probe *n* | probe 1..probe *n* |\n\nmeaning that the first *n* values (with *n* being the number of probes) are calculated using property 1 (partial charges), then another *n* values (*n*+1 up to 2*n*) calculated using property 2 (lipophilicities), and so forth.\n\n\n## 4. Release updates\n- 1.3.0:\n - Added the mode() method to the SpectrophoreCalculator class, with 'classic' and 'full' as options\n - Added the calculate_string() method to the SpectrophoreCalculator class\n - Added CI using GitHub Actions\n - Updated setup.py for easier install using pip\n - Moved the main function to a seperate main.py, making the standalone functionality more easily accessible using the spectrophore alias\n - Moved the Jupyter notebook to the main repo and updated it to the new code\n - Updated readme\n - General code cleanup\n\n- 1.2.0:\n - Switched to numpy.float32 type to achieve major speedup\n\n- 1.1.0:\n - Updated and optimised the NumPy code\n - Bug fixes\n - Introduced Numba\n - Made the 'all' normalization method the default one\n - Added a test suite \n\n- 1.0.1: First official release on PyPi\n \n\n\n## 5. Reference and citation\n\nIf you use the `spectrophore` technology in your own research work, please cite as follows:\n\nGladysz, R.; Mendes Dos Santos, F.; Langenaeker, W.; Thijs, G.; Augustyns, K.; De Winter, H. (2018) 'Spectrophores as one-dimensional descriptors calculated from three-dimensional atomic properties: applications ranging from scaffold hopping to multi-target virtual screening', *J. Cheminformatics* **10**, 9.\n",
"bugtrack_url": null,
"license": null,
"summary": "Python implementation of the spectrophore descriptor",
"version": "1.3.0",
"project_urls": {
"Download": "https://github.com/silicos-it/spectrophore/releases/latest",
"Homepage": "https://github.com/UAMCAntwerpen/spectrophore"
},
"split_keywords": [
"uamc",
" spectrophore",
" rdkit",
" cheminformatics"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "9e63464daa14f7bdf4ac7a89af91f328e423906c5b7a67a0f94df7f221c1e39b",
"md5": "7639dc9088916946d6890503acc78a38",
"sha256": "5145b57de7880bc4c3300127270f098bf28c3d74b9a774c036a7d42ceaee8098"
},
"downloads": -1,
"filename": "uamc_spectrophore-1.3.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "7639dc9088916946d6890503acc78a38",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.9",
"size": 30474,
"upload_time": "2024-09-19T10:11:21",
"upload_time_iso_8601": "2024-09-19T10:11:21.467132Z",
"url": "https://files.pythonhosted.org/packages/9e/63/464daa14f7bdf4ac7a89af91f328e423906c5b7a67a0f94df7f221c1e39b/uamc_spectrophore-1.3.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "7363c87d71194c934214ec0c14de3725082ea9d0b3fad38a2583750fa746ab88",
"md5": "2f28e041309d3d16bad952ee8ef9740b",
"sha256": "571f50d854bc7ce2d8abfe4e5c19784f479ff3b99e38cf0019db382cc7766414"
},
"downloads": -1,
"filename": "uamc_spectrophore-1.3.0.tar.gz",
"has_sig": false,
"md5_digest": "2f28e041309d3d16bad952ee8ef9740b",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.9",
"size": 40502,
"upload_time": "2024-09-19T10:11:23",
"upload_time_iso_8601": "2024-09-19T10:11:23.926438Z",
"url": "https://files.pythonhosted.org/packages/73/63/c87d71194c934214ec0c14de3725082ea9d0b3fad38a2583750fa746ab88/uamc_spectrophore-1.3.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-09-19 10:11:23",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "UAMCAntwerpen",
"github_project": "spectrophore",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"lcname": "uamc-spectrophore"
}
JSON
