Configuring your own map layer

The most important aspects of the map layer configuration are:

  • Choosing a dataset (or "layer source"), which samples will be displayed in the same layer.
  • Choosing an attribute of the dataset samples to visualize. Note: Different dataset sources have different set of attributes, e.g. EMPOP has only mt-DNA haplogroup, while AADR ancient samples has the most comprehensive set of archeological and genomical attributes.
  • Choosing a map layer type. Heatmap and points layer types are suitable for presenting a plain distribution of the samples. Especially heatmap can be used as a background of other more sophisticated visualizations. Piecharts and tags clouds can present a distribution of the given attribute values. While piecharts might be good for displaying a few chosen values, tags cloud comes in handy when all different values must be analyzed at once.

You may notice, that choosing different attributes to display results in the different set of samples being shown on the map. This is because not all samples have complete information about every attribute. If you want to see which samples have missing information, you may head to the "Features filtering" option on the map layer settings and filter the "missing" special attribute.

If you just want to display distribution of all samples from the given dataset using heatmap or plot layer type, you can choose "Dataset ID" attribute for the display, as every sample has one.

The rest of map layer options has been described at the documentation page. However, here are some more tips, that can help you configure your map layer:

  • Data weight settings are very important, as they control how the distribution is presented, where it maximizes and minimizes. "Count normalization method" relates only to the cluster of samples of the same layer. That means, if you want to normalize cluster samples by their maximal count, it will be done only in relation to the samples count of the biggest cluster in the same layer. If you want to normalize the weights by the same factor for all map layers, you should disable the "count normalization method" by choosing "None" option and use only the "weight function".
  • Effects of the map's "attributes filtering" and map layer's "features filtering" options are combined. It means that if a sample with a given attribute value is filtered out in the one the filtering options, it will not be displayed on the map. The samples with missing values are filtered out by default and it can be changed.
  • Each of the map layer types have their own set of unique options. These options affect only the specific map layer type.
  • Other grouping may help you to declutter the piecharts of the many low-count attriute values, as they will be grouped under the name of "other". The "other" group can be filtered out in the filtering settings either map's or layer's ones. The effects of "Group below frequency" and "Group below count" are combined.

If you wish to explore different map layer configuration examples you can use one of the map layer presets, which are available through the "Import predefined session" option in the "Session management" section of "Data management" panel. You can also go through the steps of the any webserver example available at the examples page to find out how each of the presented map layers was created.

Analyzing admixture results

The admixture analysis attributes (Admixture K=3, Admixture K=7 etc.) are a special type of attributes - components values for every sample must be summed to 1.0. Moreover, they have a unique set of filtering options available only for this attribute type. If you head on to the "map filtering" and "attributes filtering" section and you choose one of the admixture attributes, you may notice that the samples can be filtered by components fraction.

This comes in handy, if you want to localize the position of the certain components maximization. You can for example choose to filter only these samples, which have at least 80% of the component #3. This will show you, where are localized samples which are enriched with this component. In this way it is possible to explain the component's spatio-temporal genesis.

The same analysis as described above can be performed by querying for specific dataset samples. You can use the query builder in the "Map management" sidebar and "Database queries" section to create a filter for a specific admixture analysis and component.

Admixture components can be filtered out by name in either of map attribute filtering or map layer feature filtering options. However, it will only result in the absence of the given component for example on the piecharts. If you want to filter ut samples by the component value, you should use the aformentioned options.

Temporal grouping of samples

For now, temporal grouping of the samples is achieveable only by querying for different datasets originating from different time periods. Here is a step-by-step guide how to do it:

  • Open up the query builder in the "Map management" sidebar and "Database queries" section.
  • Now you may:
      Choose to filter the samples by an archeological period. You have to select the archeological periods name from the list.
      Choose to filter the samples by the dating years span.
      You can actually create both filters, if you want to divide specific archeological period into the early, mid and late phases.
  • Query for the dataset with the created filters. When the dataset appears in the "Loaded datasets" section, you may want to investigate it and verify, that the dating range of the samples corresponds roughly to what you wanted to achieve.
  • Repeat the process how many times you want with different archeological periods or dating time intervals.
  • Now you have to create separate map layers for each of the dataset. Configure your first map layer as described in the map layer tutorial and clone it as many times as there are datasets you want to present. Now you have to change the "layer sources" for each of them and perhaps adjust their names, style, and position, so they will not overlap and it will be clear which piecharts represent what information.
  • Optionally, it could be a good idea to cluster the samples by a region and place them in the regions centroid.

If some of the clusters still overlap, you can move them manually by dragging.

Temporal grouping was also covered in the one of our examples.

Attributes grouping options

Attributes grouping options are available in "Map grouping section" of the "Map management" sidebar. There are two different types of the attribute grouping:

  • All attribute values can be grouped by their names. To do this, you have to either manually select the values from the attribute values browser (you can make it easier using shift and ctrl keys) or select them by using searcher (you can choose to use regular expressions) and "Select all" button (it selects all visible values in the browser). With values selected, you can create a new group or add the selected values to the already exisitng group by using the "Add values to group" option.
  • For the attributes of the tree type (that means all haplogroups), you can use the automatic clustering options. These utilize the tree structure of the haplogroup data and enable to cluster all haplogroup values to the group named after their ancestor at the given tree level.

All groups can be modified, by switching the browser view to the groups. There, you can choose a group to modify, and it can be modified by excluding some of the values from the group.

It is worth to mention, that all created groups can be filtered in the attributes filtering options.

Attributes decluttering

Attributes decluttering is one of the biggest issues of the archeogenomic data presentation that we wanted to resolve with Human AGEs webserver. Below, we have described multiple strategies to declutter your visualization and to make it more readible:

  • Using the tags clouds. Because of the limited piecharts size, it is often difficult to present for example all mt-DNA haplogroups at the same time. As the tag cloud takes up more space on the map, it is more convinient to use it for presentation of the "busy" data with many different values.
  • Some of the data values might have a very low frequency but they still take a lot space in the visualization. You can filter them out by first adjusting the "Other grouping" settings in the map layer's panel and by filtering out the "other" specia attribute in the either "Map filtering" options or map layer's "Features filtering" options.
  • You can limit your visual analysis to only the specific attribute values. To do this you can use either "Attribute filtering" options in the "Map filtering" section or "Features filtering" options in the map layer's panel. Using the provided settings, you can for example filter only the descendants of the "R" Y-DNA haplogroup.
  • If data filtering is not an option, or if you desire to look at the data from a more general perspective, you may consider to perform attributes grouping. You can then choose to manualy select some attribute values and assign them to custom-name groups. It has been described in more detail in the previous tutorial.
  • If the overlapping samples clusters is you concern, you can head to the "Map grouping" and "Spatial clustering" section and adjust the "Cluster object position" value in the "Distance clustering" panel. This option controls where to display the cluster: in it's origin or in the whole clusters geometrical centroid. To declutter the clusters, to should position them in the "origin". The clusters can be always manually declutterd by dragging them.

Using interactive plots

Because the visual analysis of the PCA and UMAP plots may be difficult due to the large number of different populations, we have provided some tips that may make this task easier:

  • To explore more crowded regions of the plot, you can use the "zoom" option in the plot's top menu (the menu appears of the mouse hover). When choosed, you can draw a rectangle over any plots region to perform a zoom. To restore the original plot zoom or view, you can double-click the mouse on the plot's empty region.
  • If you want to filter out some of the population to declutter the plot, you can click their names in the legend. Additionaly, if you want to see only a specific population, you can double click it's name in the legend.

Attributes filtering options

The attribute filtering options can be performed both globally (for all layers at the same time) and individually for each of the map layers. The effect of the filters is combined.

The dataset samples can be filtered by multiple map filters at the same time. Here we described how it can be performed:

  • You have to select an attribute name from the list in the "Attributes grouing" section which is in the "Map filtering" section of the "Map management" sidebar.
  • In order to make the filter have an effect, you need to check the "is filtering active" option.
  • Now you can adjust your filtering options for each of the attributes. It is worth to mention, that each of the data type ("property" for e.g. development stage, "tree" for haplogroups and "proportion" for the admixture results) have different set of filtering options.
    • The property filtering options make it possible to filter attribute values by names. You can do it by double-clicking the names in the attribute values browser or by using other provided settings.
    • The tree filtering options for haplogroups enable to filter ancestors of descendants of the specific haplogroup in the tree.
    • The proportion filtering options for admixture results anable to filter samples if their components values meet specified condition.
  • If many filters of multiple attributes are active at the same time, their effects is combined. This means that it is possible to filter samples both by their admixture components values, development stage and Y-DNA haplogroup.

The samples presented by the given map layer can be filtered by using "Features filtering" options. It is a simplified version of the global filtering settings and it take effect only on the attribute displayed by the layer. This filtering setting is useful for example for creation of multiple heatmap, with each of them focusing on the distribution of a different haplogroup.

An example of attribute filtering was also covered here.

How to filter samples by regions

For now, the regions filtering option in the "Map management" sidebar and "Map filtering" section take effect only when the regions are displayed. The regions can be displayed only when clustering samples by regions. Therefore, to filter regions or filter samples by regions, you should first enable clutering set of regions. The different regions set filtering options are not combined - it means, that is not possible to filter samples both by archeological cultures regions and UN regions.

However, more advance options of filtering samples by regions are available through our query builder. There, you can choose to query for samples, that lay in the boundaries of many specific regions at the same time.