REFEATURE(1)

💖 REFEATURE(1) 💖

okay bestie, welcome to the refeature page from the User Commands manual ✨ she's a little command-line queen and honestly? iconic.

🌸 NAME 🌸

refeature — Read a file representing a set of features, optionally

(yes the description literally just trails off like that, no cap, that's what it says 💀)

🦋 SYNOPSIS 🦋

refeature [OPTIONS] <infile>

✨ DESCRIPTION ✨

So here's the vibe: you hand refeature a file that represents a set of features, and it will read it. Then, if you want, it can alter the set in one or more of several possible ways. Then it outputs it in whatever format you asked for. Periodt. 💅

Input and output formats may be GFF, BED, or genepred.

⚠️ okay important thing: the input format gets recognized automatically, but autorecognition needs a 'seekable' input stream — like, an actual file, not a pipe from stdin. So don't try to pipe it in, she can't rewind, she's not a VHS 🫶

👑 OPTIONS 👑

• --include-only, -i <types> — bestie this is the gatekeeper option. Include only features of the specified types (it's a comma-delimited list) and filter out literally everything else. main character only 🔥
• --include-groups, -l <file> — include only groups whose names are listed in the specified file. The group names in that file have to be delimited by white-space (and they can be spread over any number of lines, she's flexible like that ✨).
• --sort, -s — sorts features primarily by start position and secondarily by end position. Usually gives you the desired effect when you've got short overlapping features, like start & stop codons 👀. Features get sorted both across groups and within groups, but members of a group stay kept together. no splitting up the friend group fr fr 🫂
• --unique, -u — makes sure the output contains no overlapping groups (or subgroups, if you're using -e). If groups do overlap, the one with the highest score (if available) or the longest length (if there's no score) gets kept, and the others get discarded.

⚠️ warning warning: long UTRs can have undesirable results, so filter out UTR exons to avoid that mess. consider yourself warned 💀

• --groupby, -g <tag> — group features according to the specified tag. Default is "transcript_id" 💖
• --exongroup, -e <tag> — sub-group features into contiguous sets, and define those sub-groups using the specified tag (like "exon_id"). You can use this to group the features describing individual exons — e.g., each CDS and its flanking splice sites. Only features in the same major group will end up in the same minor group (so like, exons of the same transcript, ykwim? 💅).
• --fix-start-stop, -f — ensures that CDS features include start codons and exclude stop codons, as required by the GTF2 standard. She assumes at most one start_codon and at most one stop_codon per group. don't test her on that 🎀
• --add-utrs, -U — creates UTR features for the portions of exons outside CDS. Only useful with GFF output, and features must be grouped at the level of transcript 🌸
• --add-introns, -I — creates intron features between exons. Also only useful with GFF output, and features must be grouped at the level of transcript 🦋
• --add-signals, -S — adds features for start and stop codons and 3' and 5' splice sites. Only useful with GFF output, and features must be grouped at the level of transcript. Start and stop codons get added as required by the GTF2 standard, so you don't need --fix-start-stop here.

⚠️ but heads up: this does not correctly handle the case of a splice site in the middle of a start or stop codon. she said what she said 🙈

• --output, -o gff|bed|genepred|wig — output format, default is gff. And note that wig output is fixedStep and can only be used if all elements have a score and are of equal length. very specific queen energy 👑
• --simplebed, -b — for use with --output bed. Creates a separate line for each feature in bed output (by default, all features of a group are described by a single line). one line per feature, she's spreading out ✨
• --discards, -d <fname> — writes any discarded features to the specified file. like a lil lost-and-found for the features that didn't make the cut 😭
• --help, -h — prints this help message. always here for you bestie 🫶

and that's the whole page, you ate that 💅

this gorgeous thing is refeature 1.4, dated May 2016 — yeah she's vintage but she still serves, mwah 💖✨🎀