HHFILTER(1)

🌸 HHFILTER — the girlies' guide to hhfilter(1) 🌸

okay bestie, sit down, hydrate, because we're about to filter an alignment by maximum sequence identity of match states and minimum coverage and honestly? it's giving glow-up. 💅

---

✨ SYNOPSIS ✨

hhfilter -i infile -o outfile [options]

that's it. that's the whole vibe. in, out, options. 🫶

---

💖 DESCRIPTION 💖

HHfilter 3.3.0</purl> — say it with me — filters an alignment by maximum pairwise sequence identity, minimum coverage, minimum sequence identity, or score per column to the first (seed) sequence. n(c) no cap. it literally just cleans up your MSA so it's not a chaotic mess of 40,000 near-identical sequences. main character behaviour.

this iconic little tool is brought to you by the HH-suite development team, aka Steinegger M, Meier M, Mirdita M, Vöhringer H, Haunsberger S J, and Söding J (2019), in their absolute banger HH-suite3 for fast remote homology detection and deep protein annotation. BMC Bioinformatics, doi:10.1186/s12859-019-3019-7. we stan. 👑

so here's the core trio of flags, the ones you'll actually use, and yes they take <file> arguments because everything in life requires input:

• -i <file> — read input file in A3M/A2M or FASTA format. this is your raw material, your clay, your before-picture. 🎀
• -o <file> — write to output file in A3M format. the after-picture. so fresh, so clean. ✨
• -a <file> — append to output file in A3M format. for when you're not overwriting, you're adding to the collection. 💖

not me having both -o and -a because sometimes you want a new file and sometimes you want to keep stacking. the range.

---

👑 OPTIONS 👑

🌈 the tuning knobs (a.k.a. the options proper)

• -v <int> — verbose mode, and the exact levels are: 0:no screen output 1:only warings 2: verbose. so 0 = silent treatment, 1 = only the drama (warnings), 2 = tell me EVERYTHING. 💀
• -id — range [0,100], maximum pairwise sequence identity (%) (def=90). the classic. this is the one everybody uses. 💅
• -diff — range [0,inf[, filter MSA by selecting most diverse set of sequences, keeping at least this many seqs in each MSA block of length 50 (def=0). diversity hire but make it protein sequences. 🌸
• -cov — range [0,100], minimum coverage with query (%) (def=0). how much of the query you gotta actually cover to make the cut. 🦋
• -qid — range [0,100], minimum sequence identity with query (%) (def=0). same energy as -id but measured against the query specifically. 🧚
• -qsc — range [0,100], minimum score per column with query (def=-20.0). yes the range says [0,100] and the default is -20.0, i don't make the rules, i just report them. 😭
• -neff — range [1,inf[, target diversity of alignment (default=off). it's off by default, so it's lowkey asleep until you wake it up. 🔥

🎀 Input alignment format (pick your fighter)

• -M a2m — use A2M/A3M (default): upper case = Match; lower case = Insert; '-' = Delete; '.' = gaps aligned to inserts (may be omitted). the moment you learn uppercase vs lowercase actually means something. ate. ⭐
• -M first — use FASTA: columns with residue in 1st sequence are match states. the first sequence is the blueprint, the blueprint is the moment. 👑
• -M [0,100] — use FASTA: columns with fewer than X% gaps are match states. you pick the threshold, you set the standard, you are the standard. 💜

🫶 Other options (the safety rails)

• -maxseq <int> — max number of input rows (def=65535). it's giving "please don't bring 65,536 friends to my party". 🙈
• -maxres <int> — max number of HMM columns (def=20001). width limit, bestie. respect it. 🍓

---

💫 EXAMPLE 💫

because a demo eats every time:

hhfilter -id 50 -i d1mvfd_.a2m -o d1mvfd_.fil.a2m

read d1mvfd_.a2m, cap the pairwise identity at 50%, write the glow-up to d1mvfd_.fil.a2m. iconic. periodt. ✨

---

reminder to my beautiful readers: the ranges in the original are written with those fancy open brackets like [0,inf[, and i kept them exactly like that because i'm not about to misquote the source material. iykyk. 💅

---

and that's the page, bestie — this was hhfilter from the HH-suite, version 3.3.0+ds, dated August 2023. go forth, filter responsibly, and remember: no cap, only coverage. 💖🌸✨