search

kunalkathuria / SVXplorer

Structural Variant Caller
MIT License
9 stars 2 forks source link

Check if deletion thresholds etc. are best values #71

Closed kunalkathuria closed 6 years ago

kunalkathuria commented 6 years ago

Notes: Real-data INS are fewer than DELS but may look exactly like DELS in SR SRR3 PE clustering can be better? SPLIT_INS makes sense to put or not?

kunalkathuria commented 6 years ago

Checked: SRR3 gains for other tools seem to be based on SRs only, PE discordants almts are non-existent in most of these call regions

kunalkathuria commented 6 years ago

Rerun SRR5 with old stringent SPLIT_INS criteria and compare. Try DEL_T = .7 for all and compare. DUP_T = 1.3 then by "symmetry."