Microbial Land Scape

Contact our Customer Care Team:
Freecall 1300 247 301
Phone +61 3 9321 3700
CustomerCare@agrf.org.au
www.agrf.org.au

Color by:

Number of Neighbors

Local Connectivity

Minimum distance

Placeholder for UMAP figure:

The UMAP figure here will be generated by using ONLY samples uploaded on the first tab [Prediction] or the default built-in testing dataset after clicking [Predict with Random forest]. The UMAP plot reflects actual bacterial abundance within each sample of the batch. Predicted values are not used as variables in the plot but only for coloring purposes as metadata.

The parameters on the left control the shape of the UMAP plot. These parameters are arbitrarily chosen and are likely to have the most significant effect on the embedding shape.

* Color by:
that column from the predicted metadata. Note that column will be used for rain cloud plot when you select dots on UMAP with lasso tool.

* UMAP: Number of Neighbors:
control local structure: how many samples sample considered per cluster. Integer must be more than 1. This usually effect how many main sample groups you can resolve with bacterial profiles.

* UMAP: Local Connectivity:
how many connection each sample must have: minimum 1. That parameters usually linked with sample cohesivness and cluster area ther bigger number usually return much more spreaded structures.

* UMAP: Minimum distance:
is a value between 0 and 1. This number affects the trade-off between local detail and global structure in the resulting embedding. If you want to explore the fine structure within clusters, a lower number may be preferable. If you are interested in understanding the global arrangement of data, a higher number can help.

For an explanation of UMAP hyperparameters, refer to this blog written by Andy Coenen and Adam Pearce which you may find informative.

Color by:

Number of Neighbors

Local Connectivity

Minimum distance

Australian Microbiome integration

Australian Microbiome Initiative is an Australian Government funded project to catalogue microbiome data and accompanied metadata gathered across various regions of Australia. To learn more about protocols, metadata methods, and raw data generation, visit the official Australian Microbiome Initiative website by clicking on the picture on the left.

The bacterial profiles from AMI samples were generated using the AGRF DivPro pipeline, will be merged with either a built-in test dataset or an uploaded dataset. The combined dataset will be plotted using UMAP and linked to a geographical map, where the coordinates for the test data are predicted. In this visualization, new batch (test) samples are represented with the usual position icon (a hot air balloon shape), while AMI samples are displayed as semi-transparent circles. You can use the UMAP plot to select clusters of nearby samples to see how they align on the geographical map and gather metadata for the selected samples in the linked interactive table. You can use rectangle or lasso selection tools for this purpose.

NVIS and ALUM8 for environment labels

The National Vegetation Information System (NVIS) version 7.0 (Nov 2024) was used to assign soil samples spatially to major vegetation group labels (MVG), as defined in the extant layer referencing Australia’s 2016 conditions. Visit the official NVIS website by clicking on the picture on the left for more details. MVG are broad categories of vegetation types used to classify and map Australia's natural and modified landscapes. MVGs represent major structural and floristic groupings of vegetation, such as rainforests, woodlands, grasslands, and shrublands, based on dominant plant species and environmental characteristics. These classifications provide a consistent framework for understanding vegetation distribution and support environmental planning, conservation, and management. Covering the entire Australian continent, MVGs reflect the diversity of ecosystems across various climates and geographies, enabling a comprehensive representation of the country's vegetation patterns.

A substantial portion of the soil microbiome samples (~30-40%) from the AMI dataset fell into areas categorized by NVIS7 as MVG class 'Human-Affected Non-Native Vegetation'. These samples were further classified using spatial information from the Australian Land Use and Management Classification Version 8 (ALUM8) database (October 2016). This spatial layer provided more granular insights into soil specimens collected from agricultural and urban areas.

Number of Neighbors

Local Connectivity

Minimum distance

To make report files in this tab you have generate prediction values by clicking on predict button in the first tab and then embed you samples with AMI dataset on the third tab.

To save individual sample report in HTML file, select sample id from dropdown and click on download button, when it will appered. In opened download menue select destination folder on your local disk. Note it may take few seconds before file name for download appeared.

The individual sample report will be a portable (single HTML) file with an interactive UMAP embedding. Please note that the parameters for the exported UMAP will match those selected in the AMI embedding tab. The report will also include the top 10 most similar samples from the AMI dataset, the top 10 bacteria displayed in a sunburst plot, and the top 3 classifier results for the most probable environments where sample was taken.

Click the button below to perform a prediction using the built-in test dataset. Alternatively, you can upload your own data in a TSV file generated by the AGRF DivPro pipeline. For the upload, please use the bacterial relative abundance file from the Results folder named: silva_rel-table-7.tsv.

Select samples with [Lasso Select] or [Box Select] tool from umap figure on the right side. Hoover over top right corner of UMAP plot, then click, then pick the area on umap plot.

Selected vs unselected samples

Placeholder for UMAP figure:

The parameters on the left control the shape of the UMAP plot. These parameters are arbitrarily chosen and are likely to have the most significant effect on the embedding shape.

* Color by:
that column from the predicted metadata. Note that column will be used for rain cloud plot when you select dots on UMAP with lasso tool.

* UMAP: Number of Neighbors:
control local structure: how many samples sample considered per cluster. Integer must be more than 1. This usually effect how many main sample groups you can resolve with bacterial profiles.

* UMAP: Local Connectivity:
how many connection each sample must have: minimum 1. That parameters usually linked with sample cohesivness and cluster area ther bigger number usually return much more spreaded structures.

For an explanation of UMAP hyperparameters, refer to this blog written by Andy Coenen and Adam Pearce which you may find informative.

Click on the button below to perform UMAP embedding with AMI soil subset data

Geographical map

Australian Microbiome integration

NVIS and ALUM8 for environment labels

To make report files in this tab you have generate prediction values by clicking on predict button in the first tab and then embed you samples with AMI dataset on the third tab.

To save individual sample report in HTML file, select sample id from dropdown and click on download button, when it will appered. In opened download menue select destination folder on your local disk. Note it may take few seconds before file name for download appeared.

Click the button below to perform a prediction using the built-in test dataset. Alternatively, you can upload your own data in a TSV file generated by the AGRF DivPro pipeline. For the upload, please use the bacterial relative abundance file from the Results folder named: silva_rel-table-7.tsv.

Select samples with [Lasso Select] or [Box Select] tool from umap figure on the right side. Hoover over top right corner of UMAP plot, then click, then pick the area on umap plot.

Selected vs unselected samples

Placeholder for UMAP figure:

The parameters on the left control the shape of the UMAP plot. These parameters are arbitrarily chosen and are likely to have the most significant effect on the embedding shape.

* Color by: that column from the predicted metadata. Note that column will be used for rain cloud plot when you select dots on UMAP with lasso tool.

* UMAP: Number of Neighbors: control local structure: how many samples sample considered per cluster. Integer must be more than 1. This usually effect how many main sample groups you can resolve with bacterial profiles.

* UMAP: Local Connectivity: how many connection each sample must have: minimum 1. That parameters usually linked with sample cohesivness and cluster area ther bigger number usually return much more spreaded structures.

For an explanation of UMAP hyperparameters, refer to this blog written by Andy Coenen and Adam Pearce which you may find informative.

Click on the button below to perform UMAP embedding with AMI soil subset data

Geographical map

Australian Microbiome integration

NVIS and ALUM8 for environment labels

To make report files in this tab you have generate prediction values by clicking on predict button in the first tab and then embed you samples with AMI dataset on the third tab.

To save individual sample report in HTML file, select sample id from dropdown and click on download button, when it will appered. In opened download menue select destination folder on your local disk. Note it may take few seconds before file name for download appeared.

* Color by:
that column from the predicted metadata. Note that column will be used for rain cloud plot when you select dots on UMAP with lasso tool.

* UMAP: Number of Neighbors:
control local structure: how many samples sample considered per cluster. Integer must be more than 1. This usually effect how many main sample groups you can resolve with bacterial profiles.

* UMAP: Local Connectivity:
how many connection each sample must have: minimum 1. That parameters usually linked with sample cohesivness and cluster area ther bigger number usually return much more spreaded structures.