Correlating the TSV columns with HTML

[Steve Huang]

Hi Brent,

I have a simple question regarding sex QC.

Using a 30X CCS BAM for HG00514 (unfortunately, it's named as UnnamedSample in the bam...), I got the following TSV from somalier relate

#family_id	sample_id	paternal_id	maternal_id	sex	phenotype	original_pedigree_sex	gt_depth_mean	gt_depth_sd	depth_mean	depth_sd	ab_mean	ab_std	n_hom_ref	n_het	n_hom_alt	n_unknown	p_middling_ab	X_depth_mean	X_n	X_hom_ref	X_het	X_hom_alt	Y_depth_mean	Y_n
chs	UnnamedSample	0	0	2	0	female	26.6	10.9	26.5	11.0	0.51	0.43	4964	3826	61222472	0.010	24.57	324	115	74	135	0.00	0

In the HTML, here's the entry

var sample_data = [{"sample":"UnnamedSample","sex":"female","gt_depth_mean":26.56822516326648,"depth_mean":26.47785319834331,"ab_mean":0.5111052337888106,"pct_other_alleles":0.06902285407784831,"n_hom_ref":4964,"n_het":3826,"n_hom_alt":6122,"n_unknown":2472,"n_known":14912,"p_middling_ab":0.009966398997665014,"x_depth_mean":1.8498845085106,"x_hom_ref":115,"x_het":74,"x_hom_alt":135,"y_depth_mean":0.0}]

I'm curious how x_depth_mean is derived from the TSV entries (if it is indeed derived from there).

Thanks!

Steve

[Brent Pedersen]

Hi Steve,
in the html, the x_depth_mean is normalized to try to get females to ~2 and males to ~1. It uses this:

      "x_depth_mean": 2 * stats[i].x_dp.mean / stats[i].gtdp.mean,

where gtdp is the mean depth of genotyped sites in the autosome.

I think your columns must be off in the TSV table, probably n_unknown was combined with n_hom_alt . So then your tsv X_depth_mean would be 24.57

[Steve Huang]

Ah, yes, you are right.
Sorry for this mistake and thanks for being patient with me.

Thank you!

[Brent Pedersen]

no problem! glad you figured it out.