cmemit(1)

💖 cmemit — sample sequences from a covariance model ✨

okay so bestie, welcome to cmemit(1) from the Infernal Manual — this is the lil program that makes up sequences out of thin air using a covariance model, and honestly? it's giving mad scientist energy and i am obsessed. 🧪💅

---

🎀 NAME 🎀

cmemit — sample sequences from a covariance model

---

📜 SYNOPSIS 📜

cmemit [options] <cmfile>

---

🌸 DESCRIPTION 🌸

So what does cmemit actually do, bestie? It samples (emits) sequences from the covariance model(s) sitting in <cmfile>, and then writes those sequences to output. And why would you even want that? Because sampling sequences is lowkey useful for a whole bunch of purposes — like creating synthetic true positives for benchmarks or tests. iconic behaviour, no cap.

The default vibe is to sample ten unaligned sequences from each CM. But you can switch it up:

• with the -c option, you can emit a single majority-rule consensus sequence
• with the -a option, you can emit an alignment

Your <cmfile> may contain a whole library of CMs, and if it does, every single CM gets used in turn — no skipping, no favourites. 💖

And plot twist: <cmfile> may be '-' (a dash!), which means read this input from stdin rather than from a file. so convenient.

⚠️ okay, important lore for models with zero basepairs: for those, sequences are sampled from the profile HMM filter instead of the CM. Now, these models are gonna be nearly identical anyway (unless special options were used in cmbuild to prevent that), so using the HMM instead of the CM will not change the output in a significant way — unless the -l option is used. With -l, the HMM gets configured for equiprobable model begin and end positions, while the CM will not. so that's a real difference, fr fr. You can force cmemit to always sample from the CM with the --nohmmonly option. 💪

---

✨ OPTIONS ✨

• -h — Help; print a brief reminder of command line usage and available options. 🫶
• -o <f> — Save the synthetic sequences to file <f> rather than writing them to stdout.
• -N <n> — Generate <n> sequences. The default value for <n> is 10.
• -u — Write the generated sequences in unaligned format (FASTA). This is the default behavior, so it's the vibe you get for free.
• -a — Write the generated sequences in an aligned format (STOCKHOLM) with consensus structure annotation rather than FASTA. Other output formats are possible with the --outformat option. 👀
• -c — Predict a single majority-rule consensus sequence instead of sampling sequences from the CM's probability distribution. Here's the tea on the casing: highly conserved residues (base paired residues that score higher than 3.0 bits, or single stranded residues that score higher than 1.0 bits) are shown in UPPER CASE; others are shown in lowercase.
• -e <n> — Embed the CM emitted sequences in a larger randomly generated sequence of length <n> generated from an HMM that was trained on real genomic sequences with various GC contents (the same HMM used by cmcalibrate). You can use the --iid option to generate 25% A, C, G, and U sequence instead. The CM emitted sequence will begin at a random position within the larger sequence and will be included in its entirety unless the --u5p or --u3p options are used. When -e is used in combination with --u5p, the CM emitted sequence will always begin at position 1 of the larger sequence and will be truncated 5'. When used in combination with --u3p the CM emitted sequence will always end at position <n> of the larger sequence and will be truncated 3'.
• -l — Configure the CMs into local mode before emitting sequences. By default the model will be in global mode. In local mode, large insertions and deletions are more common than in global mode. so if you want chaos, you know where to find it. 💀

---

✂️ OPTIONS FOR TRUNCATING EMITTED SEQUENCES ✂️

• --u5p — Truncate all emitted sequences at a randomly chosen start position <n>, by only outputting residues beginning at <n>. A different start point is randomly chosen for each sequence. no two the same, periodt.
• --u3p — Truncate all emitted sequences at a randomly chosen end position <n>, by only outputting residues up to position <n>. A different end point is randomly chosen for each sequence.
• --a5p <n> — In combination with the -a option, truncate the emitted alignment at a randomly chosen start match position <n>, by only outputting alignment columns for positions after match state <n> - 1. <n> must be an integer between 0 and the consensus length of the model (which can be determined using the cmstat program). As a special case, using 0 as <n> will result in a randomly chosen start position. 🎲
• --a3p <n> — In combination with the -a option, truncate the emitted alignment at a randomly chosen end match position <n>, by only outputting alignment columns for positions before match state <n> + 1. <n> must be an integer between 1 and the consensus length of the model (which can be determined using the cmstat program). As a special case, using 0 as <n> will result in a randomly chosen end position. 🎲

---

🌈 OTHER OPTIONS 🌈

• --seed <n> — Seed the random number generator with <n>, an integer >= 0. If <n> is nonzero, stochastic sampling of sequences will be reproducible — the same command will give the same results, which is so satisfying. If <n> is 0, the random number generator is seeded arbitrarily, and stochastic samplings will vary from run to run of the same command. The default seed is 0.
• --iid — With -e, generate the larger sequences as 25% each A, C, G and U.
• --rna — Specify that the emitted sequences be output as RNA sequences. This is true by default. 👑
• --dna — Specify that the emitted sequences be output as DNA sequences. By default, the output alphabet is RNA.
• --idx <n> — Specify that the emitted sequences be named starting with <modelname>.<n>. By default <n> is 1.
• --outformat <s> — With -a, specify the output alignment format as <s>. Acceptable formats are: Pfam, AFA, A2M, Clustal, and Phylip. AFA is aligned fasta. Only Pfam and Stockholm alignment formats will include consensus structure annotation. ✨
• --tfile <f> — Dump tabular sequence parsetrees (tracebacks) for each emitted sequence to file <f>. Primarily useful for debugging, bestie. 🔍
• --exp <x> — Exponentiate the emission and transition probabilities of the CM by <x> and then renormalize those distributions before emitting sequences. This option changes the CM probability distribution of parsetrees relative to default. With <x> less than 1.0 the emitted sequences will tend to have lower bit scores upon alignment to the CM. With <x> greater than 1.0, the emitted sequences will tend to have higher bit scores upon alignment to the CM. This bit score difference will increase as <x> moves further away from 1.0 in either direction. If <x> equals 1.0, this option has no effect relative to default. This option is useful for generating sequences that are either more difficult (<x> < 1.0) or easier (<x> > 1.0) for the CM to distinguish as homologous from background, random sequence.
• --hmmonly — Emit from the filter profile HMM instead of the CM.
• --nohmmonly — Never emit from the filter profile HMM, always use the CM, even for models with zero basepairs. 💪

---

👀 SEE ALSO 👀

See infernal(1) for a master man page with a list of all the individual man pages for programs in the Infernal package. it's the mothership, fr fr.

For complete documentation, see the user guide that came with your Infernal distribution (Userguide.pdf); or see the Infernal web page (http://eddylab.org/infernal/).

---

©️ COPYRIGHT ©️

Copyright (C) 2023 Howard Hughes Medical Institute. Freely distributed under the BSD open source license.

For additional information on copyright and licensing, see the file called COPYRIGHT in your Infernal source distribution, or see the Infernal web page (http://eddylab.org/infernal/).

---

💌 AUTHOR 💌

http://eddylab.org

---

and that's a wrap bestie 🎀 this was cmemit(1) from Infernal 1.1.5, dated Sep 2023 — go emit some absolutely slay sequences, you've got this 💖✨🦋