brentp
Somalier relate with ONT WGS data
Hi all,
I am using Somalier on Nanopore WGS bams aligned to the T2T reference on samples from the same families. We know these samples should show relatedness, but Somalier seems to not be able to pick up the family relationships. Here is how I am running Somalier
somalier extract --sites ${SITES_PATH}/sites.chm13v2.T2T.vcf.gz \
--fasta ${REF_PATH}/t2t/hs1.fa \
"${SAMPLE}.bam"
Followed by
somalier relate --infer --groups "somalier_groups_in.tsv" "./*.somalier"
somalier relate --ped "somalier.samples.tsv" --groups "somalier_groups_in.tsv" "./*.somalier"
Here is the output I get (somalier.samples.tsv):
#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
00-004-0539_00-2948_bp 00-004-0539_00-2948_bp -9 -9 -9 -9 unknown 41.3 7.3 41.3 7.4 0.53 0.39 4121 6745 4813 1705 0.070 21.57 320 157 1 162 22.57 14
00-004-0560_00-3113_nl 00-004-0560_00-3113_nl -9 -9 -9 -9 unknown 44.4 7.8 44.4 7.8 0.53 0.39 3889 6831 4648 2016 0.077 22.46 307 154 0 153 21.29 14
00-004-0552_00-3065_nl 00-004-0549_00-2915_nl -9 -9 -9 -9 unknown 45.5 7.9 45.5 7.9 0.53 0.39 3960 6915 4709 1800 0.064 23.24 304 146 0 158 14.75 16
00-004-0552_00-3065_nl 00-004-0569_10-12454_bp -9 -9 -9 -9 unknown 55.3 8.5 55.3 8.5 0.53 0.39 3879 6886 4624 1995 0.042 55.43 315 99 110 106 0.00 0
00-004-0552_00-3065_nl 00-004-0547_00-2916_nl -9 -9 -9 -9 unknown 46.2 7.9 46.2 7.9 0.52 0.39 4041 6854 4743 1746 0.062 45.40 324 88 137 99 0.00 0
00-004-0552_00-3065_nl 00-004-0552_00-3065_nl -9 -9 -9 -9 unknown 40.5 7.3 40.4 7.3 0.52 0.39 4069 6811 4801 1703 0.074 40.00 330 90 130 110 0.00 0
00-004-0556_11-12676_bp 00-004-0556_11-12676_bp -9 -9 -9 -9 unknown 41.3 7.3 41.3 7.3 0.53 0.39 3989 6688 4854 1853 0.077 40.65 312 108 108 96 0.00 0
00-004-0552_00-3065_nl 00-004-0553_00-2980_ebv -9 -9 -9 -9 unknown 54.4 10.6 54.4 10.6 0.52 0.39 3912 6964 4561 1947 0.044 50.31 315 95 123 97 0.00 0
The html then looks like this
I have tried with and without the `--groups' flag, same result
So the samples are clearly related, but it seems to not be able to classify them properly. Is this a similar issue to the one here: https://github.com/brentp/somalier/issues/126 , to do with all samples being too related to be considered parents? Is it just an issue with Nanopore data being too noisy? Or something I am missing here, any help would be greatly appreciated. Many thanks!
Hi, --infer has some rules about depth and allele balance that must be not met. From your plot, you can see a clear set of parent-child relationships (IBS0 is 0). I would send in a pedigree file with your known relationships and verify that the colors in the plot cluster together.
Another thing is that when you run --infer, it creates a new samples file (that can be used as a ped file). So first run somalier without any ped or groups file. Then re-run passing the created samples file as the --ped argument and the html colors would be updated if all constraints are met.
Thank you, we were already running it as you described, I tried again removing groups, but no luck. Given we are using noisier ONT data I imagine we are hitting Somalier restrictions. Would there be an easy way to tweak the parameters to e.g. lower thresholds on base quality restrictions or depth?