Timeline & Changelog
Timeline
2023
July 19
June 21
- Hierarchical Clustering & Dendrogram
- PCA at HC: Now PCA data is displayed when retrieving HC
- Size plot and graphs can be now edited by the user
- Dendrogram: Color squares can be now edited by the user
- Cut-off: User can set a custom cutoff value to detect the clusters
- Summary Table: A new table is displayed summarizing the content of the detected cluster and how the actual categories/classes are distributed
- Categories as Dendrogram Metadata: One or several columns can be displayed next to the dendrogram. Each column has one coloured rectangle per spectrum. Each rectangle color matches one of the actual categories/classes labelled for the spectrum. This way, it is easy to visually check the detected clusters (unsupervised) and the actual clusters (supervised categories/classes)
- Dual Tech: MALDI & FTIR
- Now you can add your preferred technology at SignUp view
- At User Profile section you can change now your preferred technology
- Filter in Study List so MALDI-TOF or FTIR can be selected
- Antibiotic Hydrolysis Analysis
- Support for Meropenem
May 19
- Premium users mode is now activated. All module and features are now unlimited for premium users.
- FT-IR data.
- FTIR Quality Control Module: First version of QC test for FTIR data.
- Inclusion of new default ranges and internal improvements in range selection with FTIR data.
- FTIR Reproducibility Module: First Version of reproducibility Analysis for FTIR data.
- Algorithms.
- Balanced Accuracy metric at confusion matrix is now calculated and displayed in confusion matrices.
- PCA: Edit size, color and shape of displayed dots.
- Analysis.
- t-test is now available in biomarker analysis as univariate analysis.
- MALDI Biomarker Analysis Module have been redesigned.
- SNP pairwise distance heatmap: First version of Single Nucleotide Polymorphism heatmap.
May 18
https://doi.org/10.1128/jcm.01751-22
April 4
January 30
- Armonisation between FTIR workflow and Clover MSDAS have been updated.
- Rubberband algorithm has been added as Baseline correction algorithm.
- New normalization algorithms:
- Vector Normalization
- Amide I Normalization
- Amide II Normalization
- Predefined ranges for absorbance mass filtering:
- Membrane lipids [3000-2800]
- Proteins [1800-1500]
- Polysaccharides [1200-900]
January 5
https://doi.org/10.1128/jcm.01110-22
2022
November 28
- Outliers Detection Module first version release. PCA Reconstruction and Spectra Correlation analysis have been implemented.
- Shapley Values plot includes now a bar plot to show the shapley values mean.
- New file size upload limit is upgrade to 300MB.
- Validation & Identification modules run now asynchronously. You can leave the screen and recover the result later.
- Hyperparameters at classification can be now choosen manually or calculated via autotune algorithm.
- ROC and Precision-Recall curves displayed up to 4 decimals for AUC values and can be downloaded in CSV format.
- Error Rate have been added as a metric.
- LightGBM have been adjusted on how hyperparameters are optimized.
- In the ROC curve table at Biomarker Analysis module, you can click now on the table rows for selecting masses. Once they are selected, a peak matrix can be automatically generated by using those masses.
September 7
Major updates
- Multitask mode: Now supervised classifiers run in asynchronous mode. Therefore, the user can leave or close the page/tab safely or work in other tasks, because the result will be there next time the classifier page is accessed.
- Performance improvements: Algorithms now work in dedicated processes.
- Random Forest: Newly optimized hyperparameters are: number of estimators, max features, max tree depth, min split size, min number of samples per leaf.
- LightGBM: Adjustments on hyperparameters optimization. Shapley feature importances optimization.
- Biomarker Analysis: In the ROC curve table, the user can click on the table rows for selecting masses. Once they are selected, a peak matrix can be automatically generated by using those masses as biomarkers.
Minor updates
- TIC normalization: is now performed using the area under the entire spectrum.
- Find Peaks Graph: The user can click on a detected peak for displaying a label with the mass value.
- Biomarker Analysis: When a row is clicked in the ROC curve table, now it is also filtered in the Peak Distribution table.
August 28
July 7
- Light Gradient Boosting Machine (LightGBM) algorithm added
- Shapley Values analysis section added in Random Forest and LightGMB algorithms
- Confidence ellipses now can be displayed at PCA Analysis
- SVG vector image format added. Now plots and charts can be downloaded as .svg
January 10
- Fourier transform infrared (FTIR) format is now supported
- The platform now allows work with infrared spectroscopy data in all modules
- FTIR Specific preprocessing steps added
2021
October 18
- ROC curves added in all modules
- Precision-Recall curves added in all modules
July 19
June 18
June 3
June 1
February 15
- Validation Module added
- Now all models can be validated easily
- Results can be downloaded as .csv
- Resulting plots can be download as images
Contact Information | Clover Bioanalytical Software
Phone Number
(34) 958 991 543
Office
- Edif. Centro de Empresas PTS.
- Av. del Conocimiento, 41
- 18016 Granada, Spain