5. Interpretation

Overview

This page guides you through interpretation of the identification results retrieved using GPID.

The Gene Parliament

The GeneParliamentID pipeline summarises all individual gene identifications in a Gene Parliament, which represents the percentage of genes supporting all competing identifications. The Gene Parliament is presented both as a table and as a figure.

Gene Parliament table

Example file Sample_1_gpid.csv:

Sample	Rank	Identification	Species_group	Support_pct	Support_count	Parliament_size	Data_checks	ID_correct_pct	ID_close_pct	ID_wrong_pct
Sample_1	1	Species G	Group B	75.82	138	182	PASSED	100	0	0
Sample_1	2	Species T	Group B	16.48	30
Sample_1	3	Species F	Group B	3.85	7
Sample_1	4	Species A	Group B	2.75	5
Sample_1	5	Species R	Group A	0.55	1
Sample_1	5	Species B	Group C	0.55	1

Definition of variables in the Gene Parliament table

• Sample: Sample identified using GeneParliamentID

• Rank: Ranking of identification, sorted by percentage of support

• Identification: Species-level identification

• Species_group: Species group to which the identification belongs (if species groups have been specified)

• Support_pct: Percentage of genes supporting the identification

• Support_count: Number of genes supporting the identification

• Parliament_size: Number of genes in the Gene Parliament

• Data_checks: Indicates whether the user-defined thresholds for 1) gene performance, 2) alignment filtering and 3) parliament size were met.

  • PASSED if all thresholds were met
  • FAILED_1 if no gene passed the threshold for gene performance. Only a minimal gene parliament table is produced in this case.
  • FAILED_2 if no gene passed the alignment filtering thresholds. Only a minimal gene parliament table is produced in this case.
  • FAILED_3 if the parliament size threshold was not passed. A gene parliament table and figure are produced, but the fields ID_correct_pct, ID_close_pct and ID_wrong_pct are left empty.

  Note that the data checks are conducted in sequence. E.g., FAILED_2 implies that at least one gene passed the gene performance threshold.

The identification supported by most genes is the top identification. For the top identification only, the probabilities of the identification being correct, close or wrong are given based on the percentage of genes supporting the identification, following criteria that were established using a test dataset (see Method calibration):

• ID_correct_pct: Probability that the top identification is correct, i.e. correct to species level

• ID_close_pct: Probability that the top identification is close, i.e. correct to species group (if provided), excluding correct identifications.

• ID_wrong_pct: Probability that the top identification is wrong, i.e. neither correct to species nor to species group

  Notes: Because ID_close_pct does not include correct identifications, ID_close_pct and ID_correct_pct need to be added to retrieve the probability that an identification is correct to species group. For example, if ID_correct_pct is 72% and ID_close_pct is 14%, the probability of the identification to be correct to species group is 72% + 14% = 86%. ID_wrong_pct equals to 100 - (ID_correct_pct + ID_close_pct).

Gene Parliament figure

For more intuitive interpretation, the Gene Parliament is also represented as a type of bar chart termed a "lollipop plot". A maximum of the 10 highest-supported identifications is shown to maintain readability.

The file is saved in three different formats to allow full flexibility for different use cases:

• .jpg: rasterised image file. Standard image format.
• .pdf: portable document format. Vectorised file, thus zoomable without losing resolution.
• .svg: vector graphics file. Allows editing of all original figure components, e.g. in Inkscape or Photoshop.

Example file Sample_1_gpid.jpg:

Interpretation of the Gene Parliament plot:

• Each identification is represented as a horizontal line
• The length of the line represents the percentage of genes supporting the identification
• To aid interpretation, the circle at the tip of the "lollipop" is coloured depending on the species group (if specified)
• Next to each lollipop tip, the following information is provided:
  • Species identification
  • Species group (if specified)
  • Number of genes supporting identification
  • Percentage of genes supporting identification
• The parliament size, i.e. the number of genes in the Gene Parliament, is displayed on top of the plot

Identification scenarios

There are three commonly encountered scenarios of identification using GeneParliamentID:

1. One single identification supported by most genes
2. Multiple competing identifications with similar support
3. Identification not possible due to insufficient data

1. One single identification supported by most genes

Characteristics:

• A clear majority of genes supports one single identification
• Data checks passed

Simplified example:

Identification	Species_group	Support_pct	Support_count	Parliament_size	Data_checks	ID_correct_pct	ID_close_pct	ID_wrong_pct
Species V	Group D	51.46	194	377	PASSED	97	3	0
Species C	Group D	12.47	47
Species R	Group Y	9.28	35
Species M	Group A	2.65	10
Species D	Group E	1.86	7

Interpretation:
The identification as Species V is supported by 194 genes (51.46%), more than four times as many genes than the second-most supported identification Species C (47 genes / 12.47%). Species V is therefore supported by a clear majority of genes.

The data checks indicate that all filtering thresholds have been passed, indicating that there is sufficient data for a reliable identification.

Based on the results of method calibration using test samples and the percentage of genes supporting the identification, the probability for the top identification to be correct to species level is estimated to be 97%, and to species Group D a further 3% (100% total). The probability of the identification to be neither correct to species nor to species group is 0%.

2. Multiple competing identifications with similar support

Characteristics:

• The support for the top identification is similar to the next-most supported identifications
• Data checks passed

Simplified example:

Identification	Species_group	Support_pct	Support_count	Parliament_size	Data_checks	ID_correct_pct	ID_close_pct	ID_wrong_pct
Species H	Group C	15.71	55	350	PASSED	36	61	3
Species Q	Group C	14.57	51
Species B	Group C	11.71	41
Species F	Group D	5.14	18
Species I	Group A	4.29	15

Interpretation:
The identification as Species H is supported by 55 genes (15.71%), only marginally more than the second-most supported identification Species Q (51 genes / 14.57%), and Species B (41 genes / 11.71%). All three species, which also belong to the same species group, Group C, are therefore plausible identifications.

The data checks indicate that all filtering thresholds have been passed, indicating that there is sufficient data for a reliable identification.

Based on the results of method calibration using test samples and the percentage of genes supporting the identification, the probability for the top identification to be correct is estimated to be 36%, but the identification to species Group C is estimated to be correct at a further 61% (97% total). The probability of the identification to be neither correct to species nor to species group is 3%.

3. Identification not possible due to insufficient data

Characteristics:

• Different identifications supported only by very few genes
• Data check failed

Simplified example:

Identification	Species_group	Support_pct	Support_count	Parliament_size	Data_checks	ID_correct_pct	ID_close_pct	ID_wrong_pct
Species C	Group K	50	2	4	FAILED_3
Species Z	Group M	25	1
Species P	Group R	25	1

Interpretation:
The identification as Species C is supported by 2 genes (50%), twice as many as the other identifications Species Z and Species P (1 gene / 25% each).

However, the data check result FAILED_3 shows that the threshold for minimum parliament size has not been passed, indicating that there is insufficient data for a reliable identification.

Because the data check has been failed, no confidence estimates for the top identification are provided.