Filter¶

Filters candidate variants called by ProDuSe based upon the following characteristics:

Minimum molecule counts: The number of molecules supporting a variant at a given position must deviate from the number expected from simple noise in order to call a variant.

Molecule type: Stronger (duplex, strong bases) molecules will be examined for support (or lack of support) for a variant first. If too many strong molecule do not support a variant when support is expected, the variant will be discarded.

(If designated) Dual strand support: Both the positive and negative strand must support a variant confidently before it is called.

Run Using¶

produse filter

or

python /path/to/ProDuSe/ProDuSe/filter_produse.py

Parameters¶

-c –config: A configuration file which can supply any of the parameters described below.

-i –input: Input VCF file, listing candidate variants and molecule counts.

-m –molecule_stats:

A tab-delineated file listing total depth and different molecule abundances across the entire capture space. Generated by snv.py.

-o –output: Path and name to use for the output VCF file.

-sb –strand_bias_threshold:

Strand bias p-value threshold. If the strand bias of a variant is below this value, the variant will be discarded.

-st –strong_base_threshold:

Minimum number of strong bases required to call a variant. Note that this threshold will increase depending on locus and overall library characteristics for each molecule type.

-wt –weak_base_threshold:

Minimum number of weak bases required to call a variant. Note that this threshold will increase depending on locus and capture space characteristics for each molecule type.

Additional Considerations¶

If the duplicate rate of a library is extremely low, variants will be called based upon weak bases. Thus, a large number of false positive calls may be generated.