※ Documentation:

Frequently Asked Questions:


1. Q: What are ZmirP and CSZ? And what's the relationship between them?

A: ZmirP (zebrafish miRNA prediction) is a zebrafish-specific algorithm for pre-miRNA prediction with 8 new and 57 previously reported sequence and structure features. And CSZ (characterization of small RNAome for zebrafish) is an integrative platform for the analysis of the high-through sequencing data. Generally, too many putative results are generated by MIREAP or miRDeep2. ZmirP algorithm, a module of CSZ, is used for further filtering potentially false positive hits.

2. Q: How to use the local packages of CSZ?

A: Please download CSZ local package from DOWNLOAD page. Then follow the user prompts to unzip and install it on your computer. A manual was prepared for users.

3. Q: How does CSZ annotate sequenced sequences?

A: The procedure of CSZ is shown in HOME page. First, we only reserved unique reads ranging from 18 to 32 nt. The short reads observed with at least three times were assumed to be potential sRNA molecules or degradation fragments of larger RNAs, and were mapped to the zebrafish reference genome using Bowtie, with only one nucleotide mismatch. Then the identified reads were subsequently mapped to miRBase, Rfam, repeat sequences, RefSeq mRNAs, and piRNABank. By this procedure, the reads were successively classified into the following categories, including miRNA, rRNA, tRNA, and snRNA/snoRNA, genomic repeat, mRNA, and piRNA. Because repeat sequences in the annotation file were pre-classified into different classes, such as rRNA, tRNA, and snRNA/snoRNA, these RNAs were removed from the repeat sequences and recalled back to the four groups. Also, because piRNAs can locate in repetitive sequences, we further identified potential piRNAs by mapping other repetitive sequences to piRNABank.

4. Q: How does CSZ predict novel microRNAs?

A: As shown in the procedure of CSZ, unknown sequences that could not be assigned to any of known categories, are used to detect potentially novel miRNAs using MIREAP and miRDeep2 with the default settings. Because too many putative results were generated by MIREAP or miRDeep2, ZmirP algorithm is adopted for further filtering potentially false positive hits.

5. Q: Can ZmirP be used independently for pre-miRNA prediction?

A: Yes it can. ZmirP is part of the CSZ, but can be used independently. Please follow the manual to implement it.

6. Q: I have a few questions which are not listed above, how can I contact the authors of CSZ?

A: Please contact the major authors: Dr. Yu Xue and Yuangen Yao for details.



The leave-one-out (LOO) validation and 4-, 6-, 8-, 10-fold cross-validations were calculated to evaluate the performance and robustness of ZmirP. And the Receiver Operating Characteristic (ROC) curves were plotted. Through comparisons, ZmirP was demonstrated to be better than other approaches with the same dataset used in ZmirP and yielded high performance in terms of both sensitivity (95.64%) and specificity (98.84%).