Running the standalone program

To build model implementing hyperparameter tuning
Example: python3 hyper_param.py file_for_tuning

Input file description

file_for_tuning = file containing label (0/1), sequence (sequence containing pre-miRNAs and non-pre-miRNAs), and dot bracket ("(">"M", ".">"O", ")">"N"). All in one line separated by tab for a single instance.

Label	Sequence	Secondary Structure
1	UGAUAAACAAAGUGUGUAACAUCACCUCAUCUACAUGUGUGAUUUUUUUUUUGAAUAUAGACAACUU UUUAGUCAGAGUUUACAUGAGUUUUCACCUAAUUUGUGGUUUAAUUACACCGCAUAUUUGCCCAAUU UAGUGAGUAUAGUGAGUUUCUGUAGAGAAGCUCAUCUUAGAAUUAUUCAUGUAUUCCACUACUAAA	OOOOOOOOMMMMMMMOMMOOMMMMMOOMMMOOOONNNONNNNNOOOOOOOOOOOOOONNONNONNNN NMMMMMOMMOMOOMMMMMMMMMOMMMOOOOOOOOOMMMMMOOOOOOOONNNNNMMMMMMMMOOOOOO OONNNNNNNNOMMMMMMMMMMOOOONNNNNNNNNNOOOONNNOONNNNNNNNNOONOONNNNNNNO
0	AGAUCUACAAGAGAAGAUAAGUUUGAGGCAAAUUCGAGAUCUGGAAGCUGGUUUUCUCUUUACAAAU AACACUAACCCUACCAUCAAAUCAAGAAAGGAGGCUUUGAACAAAUAGCUUGAUUGAAGUAUGAAGU GGCUCGGUGGGCGACGAUGACGGGCGAGCUCCGGCGAGGGCCUGGGGGCCUGAGCGACGAUGACGG	OOMMMOOOMMMMMMMMMMOMMMMMOMMMMOOOOOOOONONNNOONNNNNONNNNNNNNNNOOOOOOO OOMMMMOOOMOMMMOOMMMMOMMMMMMMMMOOOONNNNOOOOOOOOOONNNNNNNNNONNNONONNN NMMMMMOMOMOOOOOOOOOONONONNNNNOOOOMMOOMMMMMOOOONNNNNOONNOOONNNOOOOO

Output file description

param.txt = Optimized hyparameters for Transformers.

Name	Value
learning_rate	0.583
activation	relu
activation2	selu
activation3	LeakyRelu
batch_size	40
embed_dim	28
epochs	20
num_heads	14
ff_dim	14
neurons	38
neurons2	12
dropout_rate	0.16
dropout_rate2	0.17
Optimizer	Adadelta

miWords.h5 = Hyperparameter optimized trained model for transformer.
miWords.sav = Hyperparameter optimized trained model for XGBoost.

Module 1: Sequence having length <= 400 bases
Example: sh M1.sh folderpath test

Output description

sequence_feature.tsv = Classification result of the sequence provided.

Sequence ID	T-Score	Result
seq_1	0.945	pre-miRNA
seq_2	0.154	non pre-miRNA
seq_3	0.389	non pre-miRNA
seq_4	0.458	non pre-miRNA
seq_5	0.875	pre-miRNA

Module 2 (A): Genomic sequence longer than 400 bases (Implementing Transformers + T-Score CNN when sRNA-seq data is not available)
Example: sh M2.sh folderpath genomic_sequence A (genomic_sequences is written as t2 in guide figure.)

Output description

merge.txt = output of module 2 (Sequence, Secondary Struture, Position wise T-Score).

S. No.	Sequence	Secondary structure	Position wise T-Score
1	GCTAGTAAATTTGTTGATTCATGCTTGTAGATGTACACACCACAGCAGATGCATGATGCATGGTTGGCATAGATAAGAATATAGAGGCAGGGGCA AAGGAAGCATGAGCTTGTGGGAGTTGTAGGAATGGATGCGTAGAGAAAGAAATGAAATCTGAACCAGA TTTGTGATTTGATTCTTTCCTATGACGTCCATTCCAATGATTTCCAGTCGCTCCTTCCTTGACCCAAACTGCTTCTTCTCTCCCCATTCCCCAAGGATGCAGCCAAATTACTCCTAA	...(((((.((.((((.......(((((..((((.((.((((..(((.........))).))))))))))..))))).....((((((( ((((..(((((((...((((...((((((((((.(((((((((((((((.(((((((.(.(((((.............. ))))).))))))))))))).)))))))))).))))))))))...)))))))))))..)))...))))))))........(((((.....))))))))).)).))))).....	[0.92965674, 0.07397337, 0.07397337, 0.07397337, 0.22203702, 0.92965674]

genomic_sequence.bed = output from T-Score CNN (bed6 format).

Sequence ID	Start	End	Name	Score	Strand
seq_Ath	387	667	abcd	60	+
seq_Ath	407	687	abcd	60	-

seq.txt = Sequence position window wise T-score of the sequence region in the window.

Sequence ID	Position	T-Score	Sequence	Secondary structure
seq_Ath	0	0.07397337	AAUUAGUCGUAUUCAGUUGUAAAUUCGUUUUUCGUGUAAUGGAGGGAGUACAGGACAAGCUGGCAAGUGGUCUUUGGAUUCCAUGAAGCCUGCUGCCGCGUACAGAAGUCACUAGUGUAGUAGUGGCA CUGAAACGACGUGUGCAUGCUGAUGCUGUCGCCGCAUCCCAUCCCAGUCCUUUUUUUUUUCGGGGACCCAUC	.....(((((.((((((..........((((((........))))))((((.((.((.((.(((..((((((....))..))))...))).))))))..))))....(((((((......)))))))) )))))))))).((.(.(((..(((((.......))))).))).)))((((((.........)))))).....
seq_Ath	1	0.07397337	AUUAGUCGUAUUCAGUUGUAAAUUCGUUUUUCGUGUAAUGGAGGGAGUACAGGACAAGCUGGCAAGUGGUCUUUGGAUUCCAUGAAGCCUGCUGCCGCGUACAGAAGUCACUAGUGUAGUAGUGGCAC UGAAACGACGUGUGCAUGCUGAUGCUGUCGCCGCAUCCCAUCCCAGUCCUUUUUUUUUUCGGGGACCCAUCC	....(((((.((((((..........((((((........))))))((((.((.((.((.(((..((((((....))..))))...))).))))))..))))....(((((((......))))))))) ))))))))).((.(.(((..(((((.......))))).))).)))((((((.........))))))......
seq_Ath	2	0.07397337	UUAGUCGUAUUCAGUUGUAAAUUCGUUUUUCGUGUAAUGGAGGGAGUACAGGACAAGCUGGCAAGUGGUCUUUGGAUUCCAUGAAGCCUGCUGCCGCGUACAGAAGUCACUAGUGUAGUAGUGGCACU GAAACGACGUGUGCAUGCUGAUGCUGUCGCCGCAUCCCAUCCCAGUCCUUUUUUUUUUCGGGGACCCAUCCC	...(((((.((((((..........((((((........))))))((((.((.((.((.(((..((((((....))..))))...))).))))))..))))....(((((((......)))))))))) )))))))).((.(.(((..(((((.......))))).))).)))((((((.........)))))).......
seq_Ath	3	0.07397337	UAGUCGUAUUCAGUUGUAAAUUCGUUUUUCGUGUAAUGGAGGGAGUACAGGACAAGCUGGCAAGUGGUCUUUGGAUUCCAUGAAGCCUGCUGCCGCGUACAGAAGUCACUAGUGUAGUAGUGGCACUG AAACGACGUGUGCAUGCUGAUGCUGUCGCCGCAUCCCAUCCCAGUCCUUUUUUUUUUCGGGGACCCAUCCCA	..(((((.((((((..........((((((........))))))((((.((.((.((.(((..((((((....))..))))...))).))))))..))))....(((((((......))))))))))) ))))))).((.(.(((..(((((.......))))).))).)))((((((.........))))))........

plot = folder containing "csv" files to construct line plot.
merge = folder containing "csv" files to construct line plot overlapping sequence.

Module 2 (B): If the user is providing sequencing read data (sRNA-seq) and also sequence length is >400 bases:

User needs to process the fastq file into bed6 format by instructions mentioned below (Note: User can select their own mapping software)

Step 1: Build Genome Index: hisat2-build genome index

Step 2: Read Mapping: hisat2 -x index -U read.fastq -S read.sam

Step 3: Convert SAM to bed6 format: sam2bed read.sam |cut -f1-6 > read.bed

If multiple condition are available then merge them into one bed file: cat *.bed >read.bed

Step 4: sh M2.sh folderpath genomic_sequence B read.bed (genomic_sequences is written as t2 in guide figure.)

Output description

read.bed = read alignment data in bed6 format (converted from SAM) (generated by user).

Sequence ID	Start	End	Name	Score	Strand
seq_Ath	167	179	abcd	60	-
seq_Ath	169	180	abcd	60	-
seq_Ath	237	249	abcd	60	-
seq_Ath	313	325	abcd	60	+
seq_Ath	535	549	abcd	60	+
seq_Ath	542	553	abcd	60	-
seq_Ath	542	553	abcd	60	-
seq_Ath	542	553	abcd	60	-
seq_Ath	542	553	abcd	60	-
seq_Ath	542	554	abcd	60	-

genomic_sequence = Fasta sequence utilized for Bimodal CNN.

genomic_sequence.bed = output from Bimodal CNN (bed6 format).

Sequence ID	Start	End	Name	Score	Strand
seq_Ath	91	371	abcd	60	+
seq_Ath	406	686	abcd	60	+
seq_Ath	435	715	abcd	60	-
seq_Ath	521	801	abcd	60	-

genomic_sequence_rpm.txt = Final result (ID, Start, End, Strand, Sequence, Secondary Struture).

Sequence ID	Start	End	Strand	Sequence	Secondary structure
seq_Ath	397	677	+	UUGUUGAUUCAUGCUUGUAGAUGUACACACCACAGCAGAUGCAUGAUGCAUGGUUGGCAUAGAUAAGAAUAUAGAGGCAGGGGCAAAGGAAGCAUGAGCUUG UGGGAGUUGUAGGAAUGGAUGCGUAGAGAAAGAAAUGAAAUCUGAACCAGAUUUGUGAUUUGAUUCUUUCCUAUGACGUCCAUUCCAAUGAUUUCCAGUCGCUCCUUCCUUGACCCAAACUGCUUCUU CUCUCCCCAUUCCCCAAGGAUGCAGCCAAAUUACUCCUAAUUUGCUCCUC	.(((((((.(((((.((((..(((..........)))..))))....))))))))))))......(((....(((((((((((..(((((((...((((...((((((((((.(((((((((((((((.(((((((. (.(((((..............))))).))))))))))))).)))))))))).))))))))))...)))))))))))..)))...))))))))..)))............((((.(((....(((((....)))))))))))).
seq_Ath	407	687	-	GGCCGACCUGGAGGAGCAAAUUAGGAGUAAUUUGGCUGCAUCCUUGGGGAAUGGGGAGAGAAGAAGCAGUUUGGGUCAAGGAAGGAGCGACUGGAAAUCAUU GGAAUGGACGUCAUAGGAAAGAAUCAAAUCACAAAUCUGGUUCAGAUUUCAUUUCUUUCUCUACGCAUCCAUUCCUACAACUCCCACAAGCUCAUGCUUCCUUUGCCCCUGCCUCUAUAUUCUUAUCU AUGCCAACCAUGCAUCAUGCAUCUGCUGUGGUGUGUACAUCUACAAGCAU	(((...(((.(((((((((((((....)))))....))).))))).))).((((((((((.(((.((((...((((.(((((((((((...(((........((((((((.((..((((((((((........ .(((((((...)))))))...))))))).))).)).))))))))........)))...))))...))))))).)))))))).)))...))))).))))))))....(((((...))))).((((((((((....)).)))).)))).