VCF_FILTER(1)

💖 VCF_FILTER(1) 💖

okay bestie, gather round, because today we're talking about vcf_filter from section 1 of the manual — the little script that filters your VCF files. it's giving "I have a very specific job and I do it beautifully." 🧚

✨ NAME ✨

vcf_filter — filter a VCF file. that's it, that's the whole vibe. 🎀

🌸 SYNOPSIS 🌸

vcf_filter [-h] [--no-short-circuit] [--no-filtered] [--output OUTPUT]
[--local-script LOCAL_SCRIPT] input filter [filter_args] [filter
[filter_args]] ...

look at that. a whole parade of filters. we love to see it. 💅

🦋 DESCRIPTION 🦋

so this script is part of PyVCF. yes, that PyVCF. the one you already know and love. no cap. ✨

👑 OPTIONS 👑

🌈 positional arguments

• input — the file you wanna process. and listen, if you pass - it reads from STDIN instead, which is lowkey iconic of it. (default: None)

💜 optional arguments

• -h, --help — shows you this help message and then dips. (default: False)
• --no-short-circuit — by default, the second any filter triggers on a site, processing for that site stops. this flag says "no, keep going." don't stop the filter processing on a site if any filter is triggered. (default: False)
• --output OUTPUT — the filename to write out to. it's [STDOUT] by default, and the default is literally this whole mood: <_io.TextIOWrapper name='<stdout>' mode='w' encoding='ANSI_X3.4-1968'> 😭 (default: <_io.TextIOWrapper name='<stdout>' mode='w' encoding='ANSI_X3.4-1968'>)
• --no-filtered — outputs only the sites that passed the filters. filtered ones? gone. blocked. unfollowed. (default: False)
• --local-script LOCAL_SCRIPT — a Python file sitting in your current working directory that holds the filter classes. (default: None)

🍓 mgq

filters out sites that have only low quality variants. and here's the thing, bestie — you can absolutely have a high site quality while still having a ton of low quality calls underneath it. this filter isn't fooled. it demands at least one call be above a threshold quality. periodt.

• --genotype-quality GENOTYPE_QUALITY — filters sites with no genotypes above this quality. (default: 50)

⭐ snp-only

chooses only SNP variants. just SNPs. nothing else invited to the party.

🔥 dps

thresholds the read depth per sample.

• --depth-per-sample DEPTH_PER_SAMPLE — the minimum required coverage in each sample. (default: 5)

🌈 avg-dps

thresholds the average read depth per sample, which it works out as read_depth / sample_count. math girlies rise. 🧚

• --avg-depth-per-sample AVG_DEPTH_PER_SAMPLE — the minimum required average coverage per sample. (default: 3)

💀 eb

okay this one is the main character. it filters sites that look like correlated sequencing errors. some sequencing technologies — notably pyrosequencing — produce mutation hotspots where there's a constant level of noise, producing some reference and some heterozygote calls. so this filter computes a Bayes Factor for each site by comparing the binomial likelihood of the observed allelic depths under:

• a model with constant error equal to the MAF.
• a model where each sample is the ploidy reported by the caller.

the test value is the log of the bayes factor. higher values are more likely to be errors. 👀

⚠️ note: this filter requires rpy2. no rpy2, no eb. sorry bestie.

• --eblr EBLR — filters sites above this error log odds ratio. (default: -10)

🫶 sq

filters low quailty sites. (yes, "quailty". we're keeping it. iconic typo behaviour, no notes. 💅)

• --site-quality SITE_QUALITY — filters sites below this quality. (default: 30)

🎀 a lil sign-off

that's the whole page, bestie. this was PyVCF version 0.6.7, dated October 2015, and it lives in section 1 of the manual under VCF_FILTER(1). go filter something. slay. 🦋💖✨