An experiment looking into the effects on mapping statistics and SNV calling with bwa kmer
-k
sizes 16-23.
Conclusions
SNV Calling
The effect of altering
bwa mem2 meme
kmer-k
sizes has neglible effects on calling all classes of SNVs (SNPts, SNPtv, INS, DEL, Indel), with a small gain seen in-k 16
. However, this gain comes at a 30% (47m vs 37m) increase in mapping time vs the default-k 19
.-k
of 15 improves calling vs the default of-k
19 by adding 1 TP, calling 16 fewer FP and 1 less FN. This improvement is likely not worth the substantially longer runtime of bwa.
Alignment Metrics
Of the 167 mapping alignment statistics calculated by alignstats, few are significantly different with changing
-k
. However, a few, such as the number of unpaired reads does decrease as-k
becomes smaller.
SV Calling Impacts?
SV calling has not been assessed here, but the sizable decrease in unpaired reads at
-k 15
may indicate SV calling would improve with lower-k
.
Methods
- The daylily framework was run on the 30x Novaseq google brain fastqs for HG002.
- bwa mem2 meme was run with
-k
values from 16 to 23.-k 19
being the default. - Deduplication was run with
doppelmark
and variant calling bydeepvariant
(see v0.5.3 daylily release for commands). - Alignment statistics calculated with
alignstats
.
Results
bwa mem2 meme
Runtimes
for -k
16 to 23
Number Unpaired Reads
for -k
16 to 23
SNV Performance w/Deepvariant
for -k
16 to 23
SNV Raw Data
|SAMPLE-bwa kmer|Target Region Size|TN|FN|TP|FP|Fscore| | — | — | — | — | — | — | — | |HG002-15| 2525727871| 2521853400| 19944| 3854527| 4148| 0.9968845797| |HG002-16| 2525727871| 2521853397| 19945| 3854529| 4177| 0.996880714| |HG002-17| 2525727871| 2521853399| 19950| 3854522| 4168| 0.996881224| |HG002-18| 2525727871| 2521853400| 19949| 3854522| 4164| 0.9968818686| |HG002-19| 2525727871| 2521853400| 19945| 3854526| 4164| 0.9968823874| |HG002-20| 2525727871| 2521853397| 19946| 3854528| 4189| 0.9968790374| |HG002-21| 2525727871| 2521853396| 19951| 3854524| 4164| 0.9968816123| |HG002-22| 2525727871| 2521853397| 19948| 3854526| 4179| 0.996880067| |HG002-23| 2525727871| 2521853397| 19948| 3854526| 4189| 0.996878778|