Evaluation of the VISAGE Basic Tool for Appearance and Ancestry Prediction Using PowerSeq Chemistry on the MiSeq FGx System

Round 1

Reviewer 1 Report

The VISAGE manuscript presented a very comprehensive study that evaluated a new panel using the Powerseq chemistry coupled with MPS technology.  I only have minor comments. 

The sentence on lines 45-47 about nuclear SNPs isn't really clear - perhaps rephrase to emphasize why nuclear SNPs can be informative for BGA.

On line 55 (and throughout), the authors mention the AmpliSeq chemistry - I would suggest specifying if the Ion AmpliSeq (for Thermo Fisher platforms) or Ampliseq for Illumina (for MiSeq style platforms) was used in reference 23 (unpublished). Further, I would mention that it is a chemistry designed by Thermo Fisher Scientific for custom amplicon design. 

In the heavily cited reference 23, AmpliSeq was used - is there a reason that AmpliSeq for Illumina wasn't used in this study?  Although the Powerseq chemistry seemed to perform quite well, was there a reason why the authors opted for the Promega chemistry instead of the AmpliSeq chemistry?

In section 3.2, drop-out typically doesn't have an (s).

Figure 1 - the figures in greyscale are hard to decipher (a and b) - I would suggest adding additional contrast for anyone printing the article in black and white.

The sentence on lines 191-193 is unclear - consider rephrasing.

One line 219, what do the authors mean by "these samples"?

In section 3.5, would it be useful to utilize a quant system that assesses degradation/inhibition?

On line 230, where do the 609 genotypes come from?  (4 samples X 153 markers = 612) Please clarify. 

On line 238, the authors should indicate that AmpliSeq is the chemistry. 

Very well done.  This is promising work. 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

The manuscript titled “Evaluation of the VISAGE Basic Tool for appearance and ancestry prediction using PowerSeq chemistry on the MiSeq FGx System” is a study that provides assay validation performance of a 153 SNP panel for prediction of EVCs and BGA. The study is sound, well-described, and the data analysis well supported. The study will add additional knowledge about how MPS and forensic applications will be applied to casework situations.

A few minor comments are provided for the authors consideration:

Line 85 – the authors should define the water (i.e. distilled, deionized, or whatever used).

Line 128 and onward – the authors mention replicates which are used for establishing correct calls. It is not apparent on how the replicates are used. Could it be only concordant between the two, could it be maximum alleles observed per locus, or some other mechanism of use? The authors should describe how the use the replicates and derive a result with the replicates.

Line 165 – the authors address stability for varying amounts of DNA. I do not dispute their finding. However, in this section only two input amounts were used. So varying is rather limiting to relative robust input amounts. One option to clarify is to say for these two input amounts. Another is to draw from the sensitivity study if possible to extend and support the varying amounts statement.

Line 186 – the approach to using homozygotes and assessing an alternate allele was described previously by Kieser et al vol 44, 2020 and thus should be cited.

Line 212 – the authors say they obtained 764 correct genotypes from the GEDNAP samples. The authors need to explain how they justify the types are correct. Are there ground truth data for these samples? Or are the authors using some other criteria? Or do the authors mean 764 genotype calls were obtained out of 765 possible genotype calls? The latter of these options may be the proper description. The authors should clarify.

Concordance section (3.6) - The authors describe well the one potential error of rs2789823, which may not be an error by the authors’ assay. It could be helpful to describe very briefly what would be the consequence if in a panel this one SNP typing was an error for prediction (assuming it was an error). A scenario example could put the error in perspective which I suspect has little or no consequence on prediction (at a practical level).

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

Dear authors,

This is a well-written paper and important work for the forensic community

I have some suggestions to make the paper even better (in my view):

1) How was the DNA in the test samples quantified? 

2) During the MPS proces, libraries are quantified and equally pooled before sequencing. Thus, dilution series should not be visible when just looking at number of reads pr. sample. However, in your Figure 1b there is a clear decrease in the number of reads/sample when decreasing the amount of input DNA. Can you explain this?

3) You have quantified the libraries. Were there enough library from all of the dilutions to dilute to 4 nM?

4) You have analysed the allel balance for heterozygotes markers in the 0.5 ng sample. I miss an analysis of allel balances for heterozygotes also for the samples in the dilution series. It is important to know at what DNA concentration, the heterozygote allele balance will fall out of 0.4-0.6-range, especially when working with biological stains.  

Suggestions for Figure 1:

Figure 1a: Please label the y axis

Figure 1b: y axis; "Number of reads", x axis; "DNA input (ng)"

FIgure 1c: y axis; Why is this not a box-plot? Y axis label should be "Number of reads"

Figure 1d: Both axis labels should be "ng DNA"

You probably use ggplot for plotting these Figures. If you increase the text size for both axis labels and legends, the figures will be much easier to look at (use theme())

Suggestions for Table 1:You dont need decimals

Author Response

Please see the attachment

Author Response File: Author Response.docx