Skip to content

Patch 1 - SMART3.0 documentation #8

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
ChungmaruQ wants to merge 3 commits into
base: master
Choose a base branch
from
Open

Patch 1 - SMART3.0 documentation #8

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
c6bb987
Create Introduction.md
ChungmaruQ May 17, 2021
a48759e
Create Usage.md
ChungmaruQ May 17, 2021
b7146a9
Create Contribution.md
ChungmaruQ May 17, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
56 changes: 56 additions & 0 deletions docs for 3.0/Contribution.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,56 @@

## Instructions for submitting NMR data to the Moliverse (the training set for SMART)
Please read all of the instructions below before you fill in the data collection files.

### Definitions
Folder = a folder on your computer for organizing files; File = an excel file in .xlsx or .csv format;

Sheet = a sheet within an excel file; Training database = the database used to train the AI;

Test dataset = the inquiry dataset awaiting the AI to generate structural hypotheses.

### Submitting structures to the Moliverse library
Thank you for contributing to the SMART training database (AKA the Moliverse Library). For this submission, we ask you to fill out some information in the metadata file (only one metadata file for all compounds being submitted), and a second file with the tabulated NMR data as described below (the NMR data for each compound in its own file).

Connection between the two files will be by a 6 digit code number. The starting letter/number combination for your submitted compounds will be assigned by the SMART operators, and then you should increment up for each ensuing compound by a value of 1 (e.g. AAA001 is the assigned code, the next compound in your series is AAA002, and so forth). Although compound replicates need to be avoided, please don’t worry as the SMART can cope with some replicates. You can continue adding more entries on the pre-existing metadata file with the already given code pattern.

In the Metadata file and the NMR Tables, only the items with an orange colored background are required; the green fields are desirable and will help us with developing more accurate and informative future versions of SMART. If for any compound, a specific data entry is not available, please leave it as blank.

As noted above, in the Metadata file, only the SMILES entry is mandatory. Only when the isomeric SMILES is not available for a compound, canonical SMILES can be entered instead. The SMILES can be found on the PubChem website for majority of published compounds. If you cannot find the SMILES on PubChem, the 3D structure in ChemDraw can be converted to SMILES by the program.

### Convert Structure to SMILES
To convert a ChemDraw structure to SMILES, you must:
1) Select the structure using the selection tool;
2) From the Edit menu, point to Copy As, and then choose SMILES;
3) Paste the string in the target cell of the excel metadata file sheet.

### Minimum information for NMR tables
For compiling the NMR tables, please follow the examples in "NMR Tables" file in this same folder as the format for your data. NOTE: proton shifts should be rounded to exactly 2 decimal digits, and 13C shifts should be rounded to exactly a single decimal digit.
For the NMR tables, the atom numbering is not needed, but just be certain that the proton shift matches the carbon shift in every row of the table.

In the NMR table files, wherever there are diastereotopic protons on a methylene carbon (i.e., CH2 with two distinct proton shifts), please add a separate entry for both the carbon and proton:

| Column A | Column B | Column C |
|----------|----------|------------------------|
| 13C | 1H | Carbon type (optional) |
| 43.2 | 3.12 | CH2 |
| 43.2 | 3.40 | CH2 |

In the NMR table files, please put 13C shifts in Column A of the sheet and proton shifts in Column B of the sheet, as shown above.
If you have any questions, please contact Chen Zhang (chz023 at-sign ucsd dot edu) or Raphael Reher (rreher at-sign ucsd dot edu)

Please download data submission package [here](https://tinyurl.com/vee67qk). After you complete, please email your package back to chz023 at-sign ucsd dot edu.

## References, Q&A

To reference the system please cite the papers listed below. To ask questions, please post on the user [forum](https://groups.google.com/forum/#!forum/smartnmr). We will respond you as soon as we can.

1. Zhang C\*, Idelbayev Y\*, Roberts N, Tao Y, Nannapaneni Y, Duggan BM, Min J, Lin EC, Gerwick EC, Cottrell GW, Gerwick WH. Small Molecule Accurate Recognition Technology (SMART) to Enhance Natural Products Research. *Scientific Reports.* 2017, **7(1)**, 14243. [DOI: 10.1038/s41598-017-13923-x](https://doi.org/10.1038/s41598-017-13923-x) *\*These authors contributed equally to this work.*

2. Reher R\*, Kim H\*, Zhang C\*, Mao HH, Wang M, Nothias LF, Caraballo-Rodriguez AM, Glukhov E, Teke B, Leao T, Alexander KL, Duggan BM, Van Everbroeck EL, Dorrestein PC, Cottrell GW, Gerwick WH. A Convolutional Neural Network-based approach for the Rapid Characterization of Molecularly Diverse Natural Products. *Journal of the American Chemical Society.* 2020, **142(9)**, 4114-4120. [DOI: 10.1021/jacs.9b13786](https://doi.org/10.1021/jacs.9b13786) *\*These authors contributed equally to this work.*

3. Coming soon...

End of Instructions.

<a href="https://smart.ucsd.edu/classic" target="_blank"><img src="https://user-images.githubusercontent.com/20175888/70386594-ecd8dc00-194e-11ea-8378-ba1929e90ae4.png" alt="SMART_LOGO" title="USE SMART" align="right" width="250" height="100" border="10" /></a>
126 changes: 126 additions & 0 deletions docs for 3.0/Introduction.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,126 @@
# Welcome to SMART Documentation

SMART 3.0 (Small Molecule Accurate Recognition Technology) is the latest AI-based NMR analysis.

SMART is a user-friendly, AI-based dereplication and analysis tool that uses 2D NMR data to rapidly associate newly isolated NPs with their known analogues. SMART has been designed to mimic the normal path of experiential learning in that additional 2D NMR spectral inputs can be used to enrich its database and improve its performance. In short, SMART aims to become an experienced associate to natural products researchers as well as other classes of organic chemists.

# How SMART3.0 is working?

![image](https://user-images.githubusercontent.com/51690359/118436403-a0421600-b695-11eb-85ed-98cca36bfa5f.png)

Figure 1. Overview of SMART 3.0. In the data construction phase, experimental 1H-13C HSQC data are converted to a constructed 1H-13C HSQC spectrum. In the structure annotation phase, the chemical fingerprints and molecular weights are predicted by SMART 3.0 and compared with structure databases to identify related NPs. In the class annotation phase, the compound class is predicted and provides support to the structure identification and annotation process.

# Overall framework

![image](https://user-images.githubusercontent.com/51690359/118436457-b2bc4f80-b695-11eb-9880-7e8ded31aeaf.png)

Figure 2. (A) The multi-task learning architecture of SMART 3.0. In the feature extraction step, the convolutional neural network extracts the features from HSQC spectra. Based on the extracted features, fully connected layers predict chemical fingerprints, molecular weights, and chemical classes. By using the predicted properties, structure annotation is performed. (B) The difference between HSQC and Edited HSQC spectra. On the Edited HSQC spectra, CH2 (methylene) correlations are shown separately from CH3/CH correlations. All of these correlations are binned and reconstructed on the image using 128 by 128 pixels.


# The performance of SMART 3.0 from the benchmark.

Testset from 3982 compounds is available on https://github.com/mwang87/SMART_NMR3/tree/master/Test_set

- Prediction performance

![image](https://user-images.githubusercontent.com/51690359/118436739-30805b00-b696-11eb-96bd-6eb583be8196.png)

Figure 3. Evaluation of the accuracy of SMART 3.0 to predict properties using a test set (n=3,982) not present in the training set. (A) Average (orange line) and median (blue line) of cosine scores between predicted and ground truth fingerprints for HSQC and Edited HSQC data input. (B) Linear regression between measured (x axis) and predicted molecular weights (y axis). (C) Confusion matrix of classification results using SMART 3.0 with HSQC data. (D) Confusion matrix of classification results using SMART 3.0 with Edited HSQC data.

- Structural identification/annotation performance

![image](https://user-images.githubusercontent.com/51690359/118436557-ded7d080-b695-11eb-9dfb-dc14823670be.png)

Figure 4. Performance evaluation of SMART 3.0 and other available tools with the same test set (n=3,982). Percentage of correctly identified structures (A) and annotated structures (B) found in the top k output of the different tools, for maximum rank k = 1, 2, …, 50. For the measurement of annotation rate, cosine score of 0.8 was set as the threshold.

- Performance in different solvent conditions

![image](https://user-images.githubusercontent.com/51690359/118436611-f44cfa80-b695-11eb-8b52-57c9ebeb888b.png)

Figure 5. Evaluation of SMART 3.0 analysis in different solvent conditions. (A) Venn diagram of the number of experimental NMR data obtained in chloroform-d and methanol-d4, respectively. (B) Experimental HSQC spectrum of neoline dissolved in chloroform-d¬ (blue) and methanol-d4 (red). (C) Identification (solid) and annotation (dashed) rates in total experimental data. (D) Identification (solid) and annotation (dashed) rates in compounds with NMR data recorded in both solvents.

# Frequently asked Questions

## Data Privacy

**Q:** Is the data I upload to analyze publicly visible?

**A:** Data you upload to analyze at SMART will remain private unless you explicitly make it public. The way to make your data public are to add individual annotated HSQC data to SMART database. See also the tab 'Contributing to SMART'.

## Analyzing Data

**Q:** Can I select more than one file at a time for each group in molecular networking?

**A:** Yes, you can drag&drop up to 8 CSV files in the analysis window. The analyses will run iteratively.

## NMR spectrometer Types (Bruker, JEOL, Varian...)

**Q:** Does SMART support data from different NMR vendors?

**A:** Yes, the only input you need for SMART analysis is the NMR table of 1H and 13C chemical shifts. These are vendor and processing software independent.

## Browser Support

We officially test on the latest Chrome browser. Other browsers, e.g. Firefox, Internet Explorer, Opera, Edge, are not officially supported but likely will not have any issues with SMART 3.0.


# Acknowledgements

We thank Advanced Chemical Design, Inc. for permission to utilize their Spectrus Processor 2017.2.1 software tool to predict HSQC spectra of various natural products. We further thank Dr. Kikuko Hayamizu for permission to utilize tabulated HSQC data for natural products from the CH-NMR-NP database. This work was supported by NIH grant GM107550 to G.W.C , P.C.D., and W.H.G. and by the Gordon and Betty Moore Foundation under grant GBMF7622 to G.W.C., P.C.D., and W.H.G.

We sincerely thank the following people who have done great work to make the idea of web-based SMART become true. You guys are awesome.
Names appear in alphabetic order.

| First | Middle | Last |
|-------------|:--------:|-------------------------|
| Kelsey | L. | Alexander |
| Nuno | | Bandeira |
| Wout | | Bittremieux |
| Antoine | R. | Blosse |
| Andrés | Mauricio | Caraballo-Rodriguez |
| Mitchell | | Christy |
| Garrison | W. | Cottrell |
| Alexander | | Dagman |
| Pieter | C. | Dorrestein |
| Brendan | M. | Duggan |
| Joseph | M. | Egan |
| Martha | | Gahl |
| Erik | C. | Gerwick |
| Lena | | Gerwick |
| William | H. | Gerwick |
| Michael | K. | Gilson |
| Jenny | | Hamer |
| You Kyong | | Han |
| Yerlan | | Idelbayev |
| Kyobin | | Kang |
| Hyunwoo | | Kim |
| Preston | B. | Landon |
| Aaron | | Landon |
| Tiago | F. | Leao |
| Ki Yong | | Lee |
| Bettina | | Lehman |
| Eugene | C. | Lin |
| Roger | G. | Linington |
| Zheng | | Long |
| Huanru | Henry | Mao |
| Jie | | Min |
| Anthony | | Mrse |
| Ben | C. | Naman |
| Yashwanth | | Nannapaneni |
| Louis-Félix | | Nothias |
| Poornav | S. | Purushothama |
| Siddarth | | Ravichandran |
| Raphael | | Reher |
| Nicholas | C. | Roberts |
| Hyeji | | Shin |
| Yiwen | | Tao |
| Yoshinori | | Uekusa |
| Ezra | L. | Van Everbroeck |
| Vishal | T. | Vasudevan |
| Mingxun | | Wang |
| Yiran | | Xu |
| Chen | | Zhang |
| Jianping | | Zhao |


<a href="https://smart.ucsd.edu/classic" target="_blank"><img src="https://user-images.githubusercontent.com/20175888/70386594-ecd8dc00-194e-11ea-8378-ba1929e90ae4.png" alt="SMART_LOGO" title="USE SMART" align="right" width="250" height="100" border="10" /></a>
Loading