tRF and piRNA Profiling - Usage Guide

Overview

Profile the non-miRNA small RNAs in a small RNA-seq library - tRNA-derived fragments (tRFs/tsRNAs), piRNAs, and rRNA/snoRNA-derived species - using MINTmap, unitas, SPORTS, and proTRAC. The governing reality is that small RNA-seq is a size cut over more than eight biochemically distinct classes, and the library kit decides what is captured: standard ligation needs a 5'-monophosphate and a 3'-OH, so 5'-OH and cyclic-phosphate species (tRNA halves, many tRFs) are silently absent unless a T4 PNK prep was used. Detection is not function: any abundant structured RNA sheds degradation fragments into the size window, so functionality must be earned through end precision, strand bias, phasing, the piRNA ping-pong signature, or AGO/PIWI loading. Multimapping over redundant tRNA and piRNA loci is endemic, which is why tRF tools report exclusive versus ambiguous counts and piRNAs are quantified at cluster level.

Prerequisites

conda install -c bioconda mintmap unitas protrac
# SPORTS1.0 and ShortStack (plants) are installed separately from their repositories
pip install numpy pandas

Quick Start

Tell your AI agent: - "Annotate every small-RNA class in my library and give me the composition" - "Quantify tRFs with MINTmap and use the exclusive counts" - "Detect piRNA clusters and test the ping-pong signature" - "Is this abundant species a real tRF or just tRNA degradation?" - "My library is TruSeq - can it even see tRNA halves?"

Example Prompts

Class Annotation

"Run unitas to annotate miRNA, piRNA, tRF, rRF, and snoRNA fractions in my library"

"My TruSeq library shows almost no tRNA halves - is that real or an assay artifact?"

tRF Quantification

"Profile tRFs with MINTmap and report exclusive versus ambiguous counts separately"

"Classify my tRFs as tRF-5, tRF-3, tRF-1, i-tRF, or tRNA-half"

piRNA Analysis

"Detect piRNA clusters with proTRAC and check for strand asymmetry and 1U bias"

"Compute the ping-pong z-score to confirm an active piRNA pathway"

"I see abundant piRNAs in a plasma sample - is that plausible?"

What the Agent Will Do

Check whether the prep could capture the target class (TruSeq vs T4 PNK / PANDORA-seq) before interpreting absence
Annotate all small-RNA classes (unitas/SPORTS) and read the class composition
Quantify tRFs at locus resolution with MINTmap, trusting exclusive counts
Detect piRNA clusters (proTRAC) and test the ping-pong signature for an active pathway
Separate processed functional species from degradation using end precision, strand bias, and phasing

Tips

Small RNA-seq is not a miRNA assay - it pools 8+ classes, and the kit sets the composition
Absence of a 5'-OH/cyclic-phosphate class (tRNA halves, many tRFs) in a TruSeq library is an assay artifact, not low expression - rescue needs T4 PNK prep
High read count is not evidence of function; require end precision, strand bias, phasing, ping-pong, or AGO/PIWI loading
tRNA and piRNA loci are redundant - use MINTmap exclusive counts and quantify piRNAs at cluster/family level
The ping-pong signature is a sharp 10-nt 5'-5' overlap with 1U (primary) and 10A (secondary) - it evidences the active SECONDARY (transposon) pathway specifically
A flat ping-pong score does not mean no piRNAs: primary piRNAs are phased (1U, Zucchini-dependent), and adult testis is >95% pachytene piRNAs that are non-ping-pong - test phasing too
Operationalize phasing as a peak in the same-strand 5'-to-5' distance histogram at the modal piRNA length (~26-28 nt); proTRAC reports it natively (it is the primary-pathway complement to the ping-pong test)
piRNAs carry a 3' 2'-O-methyl that suppresses standard adapter ligation, so low piRNA yield can be an end-chemistry artifact, not low abundance
Abundant piRNAs in somatic or plasma data are a red flag for misannotation (often tRFs or Y-RNA fragments matching piRBase by chance)
rRNA-derived fragments are the hardest class - even tiny rRNA decay yields huge counts, so demand reproducibility
For plants, use ShortStack (24-nt siRNA, phasiRNA biology), not animal miRNA tools

smrna-preprocessing - Size windows and end chemistry that determine class capture
mirdeep2-analysis - tRF/rRF stacks are miRDeep2 false positives
mirge3-analysis - Known miRNAs and a basic tRF module
differential-mirna - Count-based DE for tRF/piRNA matrices
genome-annotation/ncrna-annotation - tRNA/rRNA/snoRNA locus annotation