Variant Data Display Controls

1.5.2

A box located in the top-left corner of the Variants Tab contains various display option tools that allow you to control the display of the Variant data in the Variants Frequency Table (Figure 1‑53).

Figure 1‑53: The Variant data display control tools

1.5.2.1

The Alignment Read Type Controls

The ‘Alignment Read Type’ radio buttons allow you to select ‘Consensus’ or ‘Individual’. Consensi are a collapsed representation of multiple similar reads (see section 1.6.4.2) and have a single coverage value over their entire length. The intention of creating consensus reads is to simplify the data analysis and eliminate noise. However, there are sometimes discrepancies in read length within the consensus, making the coverage non-uniform. If a Variant is located in one of the regions of the consensus with lower actual coverage, the Variant frequencies reported with the ‘Consensus’ option can be misleading. Similarly, if a true variation is misinterpreted as noise, it might be eliminated from all the constructed consensi and a Variant of interest might go unnoticed. Looking at Variant frequencies based on individual reads rather than consensi gives more literal values. It is good to look at and compare Variant frequencies from both types of reads. If the numbers are in close agreement, they bolster one another, but if they are significantly different from each other, you may need to dig into the consensi to get a better understanding of the situation.

1.5.2.2

The Show Values Controls

The ‘Show values’ set of radio buttons control how to include orientation-specific Variant frequency data in the Table. Variants can potentially be found in reads or consensi of both orientations, but there may be some situations in which the design of your Amplicon libraries or the combination of Amplicon length and read length is such that certain regions of the Reference Sequence are only covered by reads of a single orientation, and Variants defined in those regions are likewise limited to single orientation coverage. Even when both orientations are available to scan for Variants, there can sometimes be discrepancies between Variant frequencies in one orientation versus the other.

•

Choosing the ‘Combined’ option merges the forward and reverse results together; the Sample-Variant cells in the Table each contain a single Variant frequency value. The Variant frequencies for the reads in each orientation are still calculated, however, and if the AVA software detects a significant difference between the combined value and the individual forward and reverse values, a small down-pointing triangle appears to the left of the value to alert you.

•	Choosing ‘Forward + reverse’ alters the format of the table so that the Sample-Variant cells are divided into two side-by-side sub-cells. The sub-cells show the Variant frequency values broken out by orientation (identified by arrowheads) rather than showing a single combined value.

•

Choosing the ‘All three’ option results in a more complex table format arrangement where each Sample-Variant cell gets subdivided into three sub-cells: the two side-by-side orientation-specific sub-cells are surmounted by the third, ‘Combined’ sub-cell (see Figure 1‑48, above). Again, a small down-pointing triangle appears to the left of the combined value if the AVA software detects a significant difference between the combined Variant frequency value and that of either orientation.

The ‘Show denominators’ checkbox adds read counts to the Sample-Variant cells as a number in parentheses following the Variant frequency values, in any displayed cell or sub-cells. This allows you to judge the reliability of Variant frequencies based on sample size and can also be of assistance when comparing Variant frequencies by orientation. If one orientation is much more highly represented than the other, you may choose to ignore the value from the underrepresented orientation. All these values (Variant frequencies and number of reads) can also be seen in the Mouse Tracker when the mouse is over a Sample-Variant cell of the Table.

1.5.2.3

The Min / Max Filters

The ‘Filter values’ controls allow you to set a minimum and maximum Variant frequency on which you want to focus in the Variant Frequency Table: Sample-Variant cells that do not meet the min and max filters are grayed-out, and if any rows or columns are entirely grayed-out, they get moved to bottom or right of the Table; the ‘Compact table’ option can remove grayed-out rows and columns from view in the Table (see section 1.5.2.4).

•

Clicking the ‘Change min/max %’ button opens the ‘Set min / max filter %’ window where you can set the minimum and maximum Variant frequency values (percentage) to use as filters, to 2 decimal places (Figure 1‑54). Values must be within the range ‘0.00’ to ‘100.00’, and the maximum value must be greater than or equal to the minimum value. Click ‘OK’ to accept and apply your min/max filter selections. The ‘All’ button resets the values to a minimum of 0.00 and a maximum of 100.00 (thus All Variant frequency values pass filter).

Figure 1‑54: The ‘Set min / max filter %’ window

•

The behavior of the minimum and maximum filters is modified by the ‘Apply min/max % to’ set of radio buttons, located beneath the ‘Change min/max %’ button. These controls are used to determine what set of Variant frequencies (forward, reverse, and/or combined) for each Sample-Variant cell is used when deciding if the cell survives the min/max filters.

◦

‘Forward or reverse’ causes the min/max settings to be applied only to the orientation-specific Variant frequency values. Either the forward or the reverse Variant frequency (or both) must meet both the minimum and the maximum filters for the Sample-Variant cell to survive the filters and remain in the table as a cell with a white background (rather than being grayed-out for failing the filter).

◦	‘Forward and reverse’ requires that the forward and reverse Variant frequency values both meet the minimum and maximum filters independently. If one orientation fails, the cell does not survive the filter and is grayed-out.

◦

‘Available data’ is a more sophisticated version of the ‘Forward and reverse’ option. In some cases, you may have intentionally sequenced only a single orientation, or the length of your Amplicon may be such that at the read length provided by the sequencing Run, the forward and reverse reads cannot provide double-orientation coverage in the region where your Variant is located. In those cases, you may not want to penalize a Variant for being represented by a single orientation when it was impossible for it to be represented in both. The ‘Available data’ option checks to see if there is read coverage from both orientations at the Variant position. If the coverage is all of one orientation, the min and max filters are applied to the Variant frequency value for that orientation. If coverage of the variant position comes from both orientations, both the forward and the reverse frequencies must independently survive the min and max filters as with the ‘Forward and reverse’ option.

•

The ‘Combined also’ checkbox can be used to also take the ‘Combined’ Variant frequency value into consideration when applying the min/max filters. If you have chosen the ‘Forward or reverse’ option, the ‘Combined’ frequency is treated like another value to be added to the ‘or’ logic, so if any of the three values meets the min/max criteria, the cell survives. When using ‘Combined’ with the ‘Forward and reverse’ or the ‘Available data’ options, the ‘Combined’ value becomes another value to be added to the ‘and’ logic, and all must pass the min/max filter or the cell fails and gets grayed-out.

1.5.2.4

The Variant Status Filter

The “Variant Status” filter is presented as a drop down menu, and its action is cumulative with that of the other filters, described above. The choices available are “All”, “Accepted or Putative”, “Accepted”, “Putative”, and “Rejected” (Figure 1‑55). Rows that do not meet the Variant Status criteria get grayed-out and moved to bottom of the Table. This filter can be useful as part of a workflow process for user verification of the data (section 1.5.2.7).

Figure 1‑55: The Variant Status filter drop down menu

1.5.2.5

The Compact Table Checkbox

The ‘Compact table’ checkbox is used to temporarily hide columns or rows that are entirely grayed-out due to a combination of lack of data, the min/max and Variant Status filter criteria, and individual row and column ‘ignore’ filter settings. This gives a less cluttered view of the data that is a better candidate for the ‘Save table snapshot’ or Save Table to Text’ options. You can uncheck the ‘Compact table’ option to reveal the rows/columns that have been hidden. In a large project that has accumulated many Samples and Variants, the judicious use of filters combined with the Compact table checkbox allow you to focus the table contents onto a meaningful domain, e.g. for preparing a report.

1.5.2.6

The Auto-Detected Variant Load Button

The Auto-Detected Variant Load button is the last control at the bottom of the Variant Display Control box. When you run a computation on the Project, the AVA software attempts to automatically detect potential substitution (SNP) and block deletion variations in the Samples that are processed (see section 1.4), and it creates a list of Variant definitions (without duplicating any Variants already saved in the Project). This queued list of potential Variants can be loaded into the Project by pressing the Load button.

The Load button obeys all the other filters in the Variant Display Control box except the Variant Status filter. The Min/Max filter values are inclusive, so if the Min is set to zero and the Max is set to 100, pressing the Load button would accept all the Auto-Detected Variants surviving the other filters.

The number of unloaded Variants that meet the collective filter criteria is indicated on the right-hand side of the Load button (Figure 1‑56A). If there are no Auto-Detected Variants that meet the current filter criteria in the queue (i.e. that have not yet been loaded), the button is grayed out and the text to the right of the button states this (Figure 1‑56B). If the Project was last computed with a version of the AVA software that did not support the Auto-Detected Variants feature (versions 1.1.01 and prior), the button is also grayed out and the text to the right states that the Project must be recomputed Figure 1‑56C).

Figure 1‑56: The “Variants Load” button in its 3 states: (A) Twenty-four detected Variants currently meet the Variants Tab filter settings, but are not yet loaded into the Project; (B) all Variants that currently meet the Variants tab filter settings (if any) have already been loaded into the Project; and (C) the current Project was computed using an earlier version of the AVA software that did not support Auto-Detected Variants (the Project must be recomputed with the current software in order to enable the Load button).

If the number of Variants to load is high, you can do a partial load by making the filter criteria more restrictive and then deal with the remainder later. For instance, you might set the filters to include Variants with a minimum frequency of 5% with support in both the Forward and Reverse Reads and press the Load button. This would give you the subset of the Variants most likely to hold up to scrutiny. A default Status of “Putative” is assigned to all the Auto-Detected Variants that get loaded into the Project.

Note that the filters in the Variant Display Control box select for Variants that meet the chosen criteria in some Sample, not ones that meet the criteria in all Samples. Thus, even with a minimum frequency setting of 5%, a Variant will be loaded if it appears at 5% or greater in one Sample, even if it is not observed at all in any of the other Samples.

No progress indicator is provided when loading Auto-Detected Variants. If the filters are set to liberal values such that a very large number of Variants are being loaded, the interface may appear to pause for few seconds during the loading process.

The loading of Auto-Detected Variants into the system is not permanent until you click the “Save” button for the Project. If you close your Project after a Variant load without clicking “Save”, the Auto-Detected Variants won’t be lost, but they will move back into the queue of the Load button and be available for import again when you reopen the Project. The full set of Auto-Detected Variants that don’t have the same pattern as any existing Project Variant is updated every time the Project is computed. In addition, the Load queue is maintained when the Project or the AVA application are closed, so it is not necessary to immediately load Variants after a computation completes.

1.5.2.7

Variant Discovery Workflow

The AVA software provides features that, when combined, provide the ability to manage a Discovery Workflow for identifying and evaluating meaningful variations. The key components of this process are the ability to load automatically detected Variants; the ability to easily set the Status of one or more Variants at a time via a right-click menu item with rows selected in the Variants Frequency Table of the main Variants Tab; and the ability to filter the content of the Variants Frequency Table based on Variant Status.

This constellation of features allows the main Variants tab to be the hub of operations for Discovery Workflow. One can choose to load Variants that have been automatically detected into the Project with the click of a single “Load” button on the Variants tab. If the volume of Variants that are available to load, as displayed to the right of the “Load” button, is large, Project clutter can be prevented by applying a selection on the Variants to load, via the filters associated with the Variants Frequency Table. For example, you could choose settings such as “Consensus” for “Alignment Read Type”, with a Min/Max of 5.00%/100.00% applied to “Forward and reverse” reads. This would allow you to load the subset of Auto-Detected Variants most likely to withstand scrutiny.

For Discovery Workflow purposes the status options have the following intended meanings:

•	Accepted – a Variant that is expected to be found in at least one Sample; or a Variant that has previously been found and verified by a user

•	Putative – a Variant that is known from the literature but may or may not be found in a Project Sample; or a Variant that has been automatically computed but not verified by a user

•	Rejected – a Variant that has been flagged as invalid because a user has determined that it was detected due to some type of artifact, such as from Sample processing or an alignment problem.

Despite these definitions, the AVA software simply treats the Variant Status as a tag that can be used for data filtering. Thus, you can choose to interpret the status values differently, if necessary, to better meet your needs.

Loaded Auto-Detected Variants are automatically assigned an initial Status of “Putative”. Any other Variants that are manually defined (section 1.3.2.5.2) or declared via filter selections on Global and Consensus alignments (see sections 1.6.3.3 and 1.7.3) will have defaulted to a Status of “Accepted”. Presuming that “Accepted” Variants have already been validated, one can set the “Variant Status” filter of the Variants Frequency Table to “Putative” and click on the “Compact Table” box. This causes any “Accepted” or “Rejected” Variant rows to be grayed-out and demoted to the bottom of the table. The “Compact Table” option then hides those rows so that the only visible rows are those that have a Status of “Putative”.

With a Project set up as above, one can begin validating the “Putative” Variants. Right-clicking on individual Sample-Variant intersection cells in the Table allows the use of the “Global Align” link to load the Global Align tab with the alignment of Read Consensi that cover the region of the corresponding Variant of interest. After exploring the underlying alignments and flowgrams to determine if the Variant appears to be legitimate, one can return to the Variants Tab to change the Status of the “Putative” Variant to either “Accepted” or “Rejected”. This is done via the “Variant Status” submenu that appears when right-clicking over a Sample-Variant intersection cell. Once the Status has been changed from “Putative”, the Variant row will no longer meet the “Variant Status” filter of the table and the row will be automatically hidden because of the active “Compact Table” option. Variants judged as invalid should be marked as “Rejected” rather than deleted entirely. This will both prevent the Variant from being added back to the Load queue, and prevent the automatic Variant detection mechanism from potentially re-proposing the same Variant after the completion of the next computation cycle (which would force you to re-evaluate this Variant each time Variants are loaded).

This method provides a shrinking pool of “Putative” Variants to work with. Eventually, after all Variants have been evaluated, the Table will be empty. If one starts the process with a partial Variant Load, looser filter settings for the Variants Frequency Table can be tried to see if the “Load” button indicates any additional Variants to load. For instance, one might keep the current filters but just change the “Forward and reverse” option to “Available data”. This would pick up Variants in regions of Amplicons that have coverage from reads of only a single orientation. Or one might try switching the “Alignment Read Type” from “Consensus” to “Individual” to catch any cases where variations are hidden because they were distributed over several Consensi at low enough levels that their changes were not incorporated into the Consensus sequences. By manipulating the filters in such a way, one can fine-tune a new set of “Putative” Variants to load and then load and validate them as before.

Eventually, the filters will be set to as restrictive values as one is willing to go, or to where there are no Variants left to load. To determine if there are any available Variants at all, one can choose the loosest settings with “Alignment Read Type” set to “Individual”, Min/Max to 0.00%/100.00%, with the Min/Max applied to “Forward or reverse” reads. At any point, one can reset the “Variant Status” filter to see “All” the Variants, just those “Accepted”, just those “Accepted or Putative”, or even those “Rejected”, and later switch back to the “Putative” only view to continue working through the list of Putative Variants.

Keep in mind that the automatic Variant detection does not currently report insertions or monomer deletions shorter than 3 bases (one must manually define those types of Variants if their statistics are to appear in the Variants Frequency Table). Also, remember that a Variant load and any changing of Variant Status are not permanent until the Project is saved. This can provide a useful form of “undo” if a Project is accidentally cluttered with an errant load. In such a case, simply re-open the Project without a “Save” in order to restore the Project’s previous list of Variants.