SUMACLUST(1)

💖 SUMACLUST 💖

okay bestie, so sumaclust is the manual page we're doing today — it lives in section 1 (User Commands) and it's literally a star clustering tool for genetic sequences ✨

🌟 NAME 🌟

sumaclust — star clustering of genetic sequences 🦋

🎀 SYNOPSIS 🎀

sumaclust [options] <dataset>

✨ DESCRIPTION ✨

so with next-gen sequencing popping off, we need tools that can handle millions of sequences in a reasonable amount of time, and that's exactly what this is, no cap. Sumaclust is a program developed by the LECA 💜 and its whole thing is clustering sequences in a way that's fast AND exact at the same time — like, both. icon behaviour honestly.

It was built to be adapted to the type of data that DNA metabarcoding generates, i.e. entirely sequenced, short markers 🧬. And it clusters sequences using the same clustering algorithm as UCLUST and CD-HIT. This algorithm is mainly useful for detecting the 'erroneous' sequences that get created during amplification and sequencing protocols — you know, the ones deriving from the 'true' sequences. it's giving cleanup crew 💅

💅 OPTIONS 💅

• -h — [H]elp, bestie. prints <this> help 🫶
• -l — Reference sequence length is the shortest.
• -L — Reference sequence length is the largest.
• -a — Reference sequence length is the alignment length (default).
• -n — Score is normalized by reference sequence length (default).
• -r — Raw score, not normalized.
• -d — Score is expressed in distance (default : score is expressed in similarity).
• -t ##.## — Score threshold for clustering. And okay, this one has lore, so listen up:
  • If the score is normalized and expressed in similarity (which is the default), then it's an identity — e.g. 0.95 for an identity of 95%.
  • If the score is normalized and expressed in distance, it's (1.0 - identity) — e.g. 0.05 for an identity of 95%.
  • If the score is not normalized and expressed in similarity, it's the length of the Longest Common Subsequence.
  • If the score is not normalized and expressed in distance, it's (reference length - LCS length).
  • Only sequences with a similarity above ##.## with the center sequence of a cluster get assigned to that cluster. Default: 0.97. 👑
• -e — Exact option ✨ A sequence is assigned to the cluster whose center sequence presents the highest similarity score > threshold. This is the opposite of the default 'fast' option, where a sequence is assigned to the first cluster found with a center sequence presenting a score > threshold.
• -R ## — Maximum ratio between the counts of two sequences so that the less abundant one can be considered as a variant of the more abundant one. Default: 1.0.
• -p ## — Multithreading with ## threads using openMP 🔥
• -s #### — Sorting by ####. Must be 'None' for no sorting, or a key in the fasta header of each sequence — except for the count, which can be computed (default : sorting by count).
• -o — Sorting is in ascending order (default : descending).
• -g — n's are replaced with a's (default: sequences with n's are discarded). 🙈
• -B ### — Output of the OTU table in BIOM format is activated, and written to file ###.
• -O ### — Output of the OTU map (observation map) is activated, and written to file ###.
• -F ### — Output in FASTA format is written to file ### instead of standard output.
• -f — Output in FASTA format is deactivated.

Argument : the nucleotide dataset to cluster 🧬

🌈 SEE ALSO 🌈

http://metabarcoding.org/sumatra

and that's the whole page, bestie 🎀 this was sumaclust version 1.0.10, dated August 2015 — literally a classic, we stan a vintage icon 💖✨🦋