However, not all duplicated genes seem to have arisen from true duplications, but some may have been transferred horizontally, like some PTS system genes that are L. monocytogenes EGD-e strain-specific genes.
The reviewers raise the concern that small-scale duplications (either true duplications or due to mis-assemblies) could generate false positives.
It should be emphasized that not all true duplications could be detected by qRT-PCR, especially low-copy duplications with divergent reads ratio > 0.8 are difficult to be detected.
In this hypothetical example, only four gene duplicate pairs representing the true duplication events were retained, while removing the six additional duplicate pairs displaying sequence similarity.
Filamin C (FLNC) appears to be a true duplication, with transcripts divergent at the nucleotide and protein levels.
All trees with four taxa are true duplication trees, thus evaluation of the A domains with the PDH algorithm is trivial.
We also removed shadow pairs within multiple-member gene families, which were representative of sequence similarity rather than true duplication events.
One gene, TNFRSF10A, which is listed provisionally here as artifactually duplicated, proved especially problematic since there is equal evidence for both artifactual and true duplication; directed studies will be necessary to determine the nature of this duplication.
Evidence for the true duplication of 13 immune-related genes: ATF4, CD36, CD68, CD163, CRP, DDX3X, GSTP1, GZMA, IFIT1, IL1B, IRGC, ITLN2, and OAS1, and 5 non-immune genes, appears unequivocal (Additional file 3: Table S3).
Further scrutiny of the nucleotide and amino acid alignments, showed that the PE_PGRS17 (Rv0978c) and PE_PGRS18 (Rv0980c) genes shared the highest identity, indicating that they may represent a true duplication event.
Precision is defined as the fraction of the number of true tandem duplications (true positives) among all tandem duplications identified by our algorithm (true positives and false positives).
