Subtitle

"Synthetic protein dataset with sequences, physical properties, and functional classification for machine learning tasks."

Description

Introduction

This synthetic dataset was created to explore and develop machine learning models in bioinformatics. It contains 20,000 synthetic proteins, each with an amino acid sequence, calculated physicochemical properties, and a functional classification.

Columns Included

• ID_Protein: Unique identifier for each protein.
• Sequence: String of amino acids.
• Molecular_Weight: Molecular weight calculated from the sequence.
• Isoelectric_Point: Estimated isoelectric point based on the sequence composition.
• Hydrophobicity: Average hydrophobicity calculated from the sequence.
• Total_Charge: Sum of the charges of the amino acids in the sequence.
• Polar_Proportion: Percentage of polar amino acids in the sequence.
• Nonpolar_Proportion: Percentage of nonpolar amino acids in the sequence.
• Sequence_Length: Total number of amino acids in the sequence.
• Class: The functional class of the protein, one of five categories: Enzyme, Transport, Structural, Receptor, Other.

Inspiration and Sources

While this is a simulated dataset, it was inspired by patterns observed in real protein datasets, such as: - UniProt: A comprehensive database of protein sequences and annotations. - Kyte-Doolittle Scale: Calculations of hydrophobicity. - Biopython: A tool for analyzing biological sequences.

Proposed Uses

This dataset is ideal for: - Training classification models for proteins. - Exploratory analysis of physicochemical properties of proteins. - Building machine learning pipelines in bioinformatics.

How This Dataset Was Created

1. Sequence Generation: Amino acid chains were randomly generated with lengths between 50 and 300 residues.
2. Property Calculation: Physicochemical properties were calculated using the Biopython library.
3. Class Assignment: Classes were randomly assigned for classification purposes.

Limitations

• The sequences and properties do not represent real proteins but follow patterns observed in natural proteins.
• The functional classes are simulated and do not correspond to actual biological characteristics.

Data Split

The dataset is divided into two subsets: - Training: 16,000 samples (proteinas_train.csv). - Testing: 4,000 samples (proteinas_test.csv).

Acknowledgment

This dataset was inspired by real bioinformatics challenges and designed to help researchers and developers explore machine learning applications in protein analysis.

This collection contains an example MINUTE-ChIP dataset to run minute pipeline on, provided as supporting material to help users understand the results of a MINUTE-ChIP experiment from raw data to a primary analysis that yields the relevant files for downstream analysis along with summarized QC indicators. Example primary non-demultiplexed FASTQ files provided here were used to generate GSM5493452-GSM5493463 (H3K27m3) and GSM5823907-GSM5823918 (Input), deposited on GEO with the minute pipeline all together under series GSE181241. For more information about MINUTE-ChIP, you can check the publication relevant to this dataset: Kumar, Banushree, et al. "Polycomb repressive complex 2 shields naïve human pluripotent cells from trophectoderm differentiation." Nature Cell Biology 24.6 (2022): 845-857. If you want more information about the minute pipeline, there is a public biorXiv and a GitHub repository and official documentation.

This dataset represents the https://github.com/PacificBiosciences/DevNet/wiki/8-plex-Ecoli-Multiplexed-Microbial-Assembly pacbio dataset already assembled and annotated so that users that want to skip some steps of the tutorial can do it by downloading this dataset.

This is the example dataset used in the CommPath R package tutorial. We downloaded the processed scRNA-seq data on hepatocellular carcinoma (HCC) samples from the paper of Sharma, et al., 2020, and then randomly selected the expression data of 3000 cells across the top 5000 highly variable genes from the tumor and normal tissues, respectively.

Three examples dataset to perform bioinformatics analysis.

Example datasets for AlphaCRV: A Pipeline for Identifying Accurate Binder Topologies in Mass-Modeling with AlphaFold. With these datasets you can replicate two of the three examples shown in the paper, following along with the Jupyter notebooks in the GitHub repository at https://github.com/strubelab/AlphaCRV. Description:

• AVRPia.fasta, AVRPik.fasta, SKP1.fasta: Sequences of the bait molecules for the three examples.
• AVRPia_binders.fasta, AVRPik_binders.fasta, SKP1_binders.fasta: Sequences of the binder molecules for the three examples.
• AVRPia_vs_rice_models.tar.lzma, AVRPik_vs_rice_models.tar.lzma: Compressed archives of the AlphaFold-Multimer models for the AVRPia and AVRPik examples. The 712 complexes for SKP1 are more than 100GB in size, so those can be provided upon request. To decompress the .tar.lzma archives use the following two commands on each:

unlzma AVRPia_vs_rice_models.tar.lzma
tar -xvf AVRPia_vs_rice_models.tar

• AVRPia_vs_rice_clusters.tar.gz, AVRPik_vs_rice_clusters.tar.gz, SKP1_vs_rice_clusters.tar.gz: Compressed archives with the results from running AlphaCRV on the three examples presented in the paper. To decompress these archives use the following command on each:

tar -xvzf AVRPia_vs_rice_clusters.tar.gz

This dataset contains 3,000 synthetic DNA samples with 13 features designed for genomic data analysis, machine learning, and bioinformatics research. Each row represents a unique DNA sample with both sequence-level and statistical attributes.

🔹 Dataset Structure

Rows: 3,000

Columns: 13

🔹 Features Description

1. Sample_ID → Unique identifier for each DNA sample

2. Sequence → DNA sequence (string of A, T, C, G)

3. GC_Content → Percentage of Guanine (G) and Cytosine (C) in the sequence

4. AT_Content → Percentage of Adenine (A) and Thymine (T) in the sequence

5. Sequence_Length → Total sequence length

6. Num_A → Number of Adenine bases

7. Num_T → Number of Thymine bases

8. Num_C → Number of Cytosine bases

9. Num_G → Number of Guanine bases

10. kmer_3_freq → Average 3-mer (triplet) frequency score

11. Mutation_Flag → Binary flag indicating mutation presence (0 = No, 1 = Yes)

12. Class_Label → Class of the sample (Human, Bacteria, Virus, Plant)

13. Disease_Risk → Risk level associated with the sample (Low / Medium / High)

🔹 Potential Use Cases

DNA classification tasks (e.g., predicting species from DNA sequence features)

Exploratory Data Analysis (EDA) in bioinformatics

Machine Learning model development (Logistic Regression, Random Forest, SVM, Neural Networks)

Deep Learning approaches (LSTM, CNN, Transformers for sequence learning)

Mutation detection and disease risk analysis

Teaching and practicing biological data preprocessing techniques

🔹 Why This Dataset?

Synthetic but realistic structure, inspired by genomics data

Balanced and diverse distribution of features and labels

Suitable for beginners and researchers to practice classification, visualization, and model comparison

🔹 Example Research Questions

Can we classify DNA samples into their biological class using sequence-based features?

How does GC content relate to mutation risk?

Which ML model performs best for DNA classification tasks?

Can synthetic DNA features predict disease risk categories?

📌 Acknowledgment

This dataset is synthetic and generated for educational & research purposes. It does not represent real patient data.

An example single-cell dataset in .rda format for use in the vignettes of the SingleCellHaystack on GitHub. Individual datasets included in the .rda file are also present as .csv files.

We used the human genome reference sequence in its GRCh38.p13 version in order to have a reliable source of data in which to carry out our experiments. We chose this version because it is the most recent one available in Ensemble at the moment. However, the DNA sequence by itself is not enough, the specific TSS position of each transcript is needed. In this section, we explain the steps followed to generate the final dataset. These steps are: raw data gathering, positive instances processing, negative instances generation and data splitting by chromosomes.

First, we need an interface in order to download the raw data, which is composed by every transcript sequence in the human genome. We used Ensembl release 104 (Howe et al., 2020) and its utility BioMart (Smedley et al., 2009), which allows us to get large amounts of data easily. It also enables us to select a wide variety of interesting fields, including the transcription start and end sites. After filtering instances that present null values in any relevant field, this combination of the sequence and its flanks will form our raw dataset. Once the sequences are available, we find the TSS position (given by Ensembl) and the 2 following bases to treat it as a codon. After that, 700 bases before this codon and 300 bases after it are concatenated, getting the final sequence of 1003 nucleotides that is going to be used in our models. These specific window values have been used in (Bhandari et al., 2021) and we have kept them as we find it interesting for comparison purposes. One of the most sensitive parts of this dataset is the generation of negative instances. We cannot get this kind of data in a straightforward manner, so we need to generate it synthetically. In order to get examples of negative instances, i.e. sequences that do not represent a transcript start site, we select random DNA positions inside the transcripts that do not correspond to a TSS. Once we have selected the specific position, we get 700 bases ahead and 300 bases after it as we did with the positive instances.

Regarding the positive to negative ratio, in a similar problem, but studying TIS instead of TSS (Zhang135
et al., 2017), a ratio of 10 negative instances to each positive one was found optimal. Following this136
idea, we select 10 random positions from the transcript sequence of each positive codon and label them137
as negative instances. After this process, we end up with 1,122,113 instances: 102,488 positive and 1,019,625 negative sequences. In order to validate and test our models, we need to split this dataset into three parts: train, validation and test. We have decided to make this differentiation by chromosomes, as it is done in (Perez-Rodriguez et al., 2020). Thus, we use chromosome 16 as validation because it is a good example of a chromosome with average characteristics. Then we selected samples from chromosomes 1, 3, 13, 19 and 21 to be part of the test set and used the rest of them to train our models. Every step of this process can be replicated using the scripts available in https://github.com/JoseBarbero/EnsemblTSSPrediction.

The table contains the name of the repository, the type of example (issue tracking, branch structure, unit tests), and the URL of the example. All URLs are prefixed with https://github.com/.

Open data science and algorithm development competitions offer a unique avenue for rapid discovery of better computational strategies. We highlight three examples in computational biology and bioinformatics research in which the use of competitions has yielded significant performance gains over established algorithms. These include algorithms for antibody clustering, imputing gene expression data, and querying the Connectivity Map (CMap). Performance gains are evaluated quantitatively using realistic, albeit sanitized, data sets. The solutions produced through these competitions are then examined with respect to their utility and the prospects for implementation in the field. We present the decision process and competition design considerations that lead to these successful outcomes as a model for researchers who want to use competitions and non-domain crowds as collaborators to further their research.

Example dataset for bioinformatic tutorials

“Bioinformatics: Introduction and Methods,” a Bilingual Massive Open Online Course (MOOC) as a New Example for Global Bioinformatics Education

A BioInformatics Dataset where it's possible to know about whether Lysine (residue 'k') is glycated or not. All the position of Lysine (alphabet k) given on 'Accession' is positive sites. Other Lysine(k) on every sequence is negative sites. You can consider window size as 15 (7 upstream, 'k' in middle, 7 downstream).

https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F3928176%2Feec4b7e199825d4d4b687585d61cec2b%2FCapture.PNG?generation=1600608829159271&alt=media" alt="">

Example after pre-processing : | Sequence | label | | ---| --- | |AKSHPPDKWAQGAGA | 1 | |GILPILVKCLERDDN | 1 | |PSSNPHAKPSDFHFL| 0 | |EEVFYAVKVLQKKAI| 0 |

There are around 6557 positive sites and 174023 negative sites. So predicting would be tough for an imbalance dataset. you can apply CD-HIT to reduce the number of negative sites. To extract features you can use CKSAAP or PSI-BLAST. It's a BioInformatics problem, not a general one. So it's better if you have domain knowledge or about protein structure.

Dataset link: http://plmd.biocuckoo.org/download.php

This dataset represents the https://github.com/PacificBiosciences/DevNet/wiki/8-plex-Ecoli-Multiplexed-Microbial-Assembly pacbio dataset already demultiplexed and preprocessed in fastq format so that users that want to skip some steps of the tutorial can do it by downloading this dataset.

Pathogen diversity resulting in quasispecies can enable persistence and adaptation to host defenses and therapies. However, accurate quasispecies characterization can be impeded by errors introduced during sample handling and sequencing which can require extensive optimizations to overcome. We present complete laboratory and bioinformatics workflows to overcome many of these hurdles. The Pacific Biosciences single molecule real-time platform was used to sequence PCR amplicons derived from cDNA templates tagged with universal molecular identifiers (SMRT-UMI). Optimized laboratory protocols were developed through extensive testing of different sample preparation conditions to minimize between-template recombination during PCR and the use of UMI allowed accurate template quantitation as well as removal of point mutations introduced during PCR and sequencing to produce a highly accurate consensus sequence from each template. Handling of the large datasets produced from SMRT-UMI sequencing was facilitated by a novel bioinformatic pipeline, Probabilistic Offspring Resolver for Primer IDs (PORPIDpipeline), that automatically filters and parses reads by sample, identifies and discards reads with UMIs likely created from PCR and sequencing errors, generates consensus sequences, checks for contamination within the dataset, and removes any sequence with evidence of PCR recombination or early cycle PCR errors, resulting in highly accurate sequence datasets. The optimized SMRT-UMI sequencing method presented here represents a highly adaptable and established starting point for accurate sequencing of diverse pathogens. These methods are illustrated through characterization of human immunodeficiency virus (HIV) quasispecies. Methods This serves as an overview of the analysis performed on PacBio sequence data that is summarized in Analysis Flowchart.pdf and was used as primary data for the paper by Westfall et al. "Optimized SMRT-UMI protocol produces highly accurate sequence datasets from diverse populations – application to HIV-1 quasispecies" Five different PacBio sequencing datasets were used for this analysis: M027, M2199, M1567, M004, and M005 For the datasets which were indexed (M027, M2199), CCS reads from PacBio sequencing files and the chunked_demux_config files were used as input for the chunked_demux pipeline. Each config file lists the different Index primers added during PCR to each sample. The pipeline produces one fastq file for each Index primer combination in the config. For example, in dataset M027 there were 3–4 samples using each Index combination. The fastq files from each demultiplexed read set were moved to the sUMI_dUMI_comparison pipeline fastq folder for further demultiplexing by sample and consensus generation with that pipeline. More information about the chunked_demux pipeline can be found in the README.md file on GitHub. The demultiplexed read collections from the chunked_demux pipeline or CCS read files from datasets which were not indexed (M1567, M004, M005) were each used as input for the sUMI_dUMI_comparison pipeline along with each dataset's config file. Each config file contains the primer sequences for each sample (including the sample ID block in the cDNA primer) and further demultiplexes the reads to prepare data tables summarizing all of the UMI sequences and counts for each family (tagged.tar.gz) as well as consensus sequences from each sUMI and rank 1 dUMI family (consensus.tar.gz). More information about the sUMI_dUMI_comparison pipeline can be found in the paper and the README.md file on GitHub. The consensus.tar.gz and tagged.tar.gz files were moved from sUMI_dUMI_comparison pipeline directory on the server to the Pipeline_Outputs folder in this analysis directory for each dataset and appended with the dataset name (e.g. consensus_M027.tar.gz). Also in this analysis directory is a Sample_Info_Table.csv containing information about how each of the samples was prepared, such as purification methods and number of PCRs. There are also three other folders: Sequence_Analysis, Indentifying_Recombinant_Reads, and Figures. Each has an .Rmd file with the same name inside which is used to collect, summarize, and analyze the data. All of these collections of code were written and executed in RStudio to track notes and summarize results. Sequence_Analysis.Rmd has instructions to decompress all of the consensus.tar.gz files, combine them, and create two fasta files, one with all sUMI and one with all dUMI sequences. Using these as input, two data tables were created, that summarize all sequences and read counts for each sample that pass various criteria. These are used to help create Table 2 and as input for Indentifying_Recombinant_Reads.Rmd and Figures.Rmd. Next, 2 fasta files containing all of the rank 1 dUMI sequences and the matching sUMI sequences were created. These were used as input for the python script compare_seqs.py which identifies any matched sequences that are different between sUMI and dUMI read collections. This information was also used to help create Table 2. Finally, to populate the table with the number of sequences and bases in each sequence subset of interest, different sequence collections were saved and viewed in the Geneious program. To investigate the cause of sequences where the sUMI and dUMI sequences do not match, tagged.tar.gz was decompressed and for each family with discordant sUMI and dUMI sequences the reads from the UMI1_keeping directory were aligned using geneious. Reads from dUMI families failing the 0.7 filter were also aligned in Genious. The uncompressed tagged folder was then removed to save space. These read collections contain all of the reads in a UMI1 family and still include the UMI2 sequence. By examining the alignment and specifically the UMI2 sequences, the site of the discordance and its case were identified for each family as described in the paper. These alignments were saved as "Sequence Alignments.geneious". The counts of how many families were the result of PCR recombination were used in the body of the paper. Using Identifying_Recombinant_Reads.Rmd, the dUMI_ranked.csv file from each sample was extracted from all of the tagged.tar.gz files, combined and used as input to create a single dataset containing all UMI information from all samples. This file dUMI_df.csv was used as input for Figures.Rmd. Figures.Rmd used dUMI_df.csv, sequence_counts.csv, and read_counts.csv as input to create draft figures and then individual datasets for eachFigure. These were copied into Prism software to create the final figures for the paper.

This workflow adapts the approach and parameter settings of Trans-Omics for precision Medicine (TOPMed). The RNA-seq pipeline originated from the Broad Institute. There are in total five steps in the workflow starting from:

1. Read alignment using STAR which produces aligned BAM files including the Genome BAM and Transcriptome BAM.
2. The Genome BAM file is processed using Picard MarkDuplicates producing an updated BAM file containing information on duplicate reads (such reads can indicate biased interpretation).
3. SAMtools index is then employed to generate an index for the BAM file, in preparation for the next step.
4. The indexed BAM file is processed further with RNA-SeQC which takes the BAM file, human genome reference sequence and Gene Transfer Format (GTF) file as inputs to generate transcriptome-level expression quantifications and standard quality control metrics.
5. In parallel with transcript quantification, isoform expression levels are quantified by RSEM. This step depends only on the output of the STAR tool, and additional RSEM reference sequences.

For testing and analysis, the workflow author provided example data created by down-sampling the read files of a TOPMed public access data. Chromosome 12 was extracted from the Homo Sapien Assembly 38 reference sequence and provided by the workflow authors. The required GTF and RSEM reference data files are also provided. The workflow is well-documented with a detailed set of instructions of the steps performed to down-sample the data are also provided for transparency. The availability of example input data, use of containerization for underlying software and detailed documentation are important factors in choosing this specific CWL workflow for CWLProv evaluation.

This dataset folder is a CWLProv Research Object that captures the Common Workflow Language execution provenance, see https://w3id.org/cwl/prov/0.5.0 or use https://pypi.org/project/cwl

Biological data is increasing at a high speed, creating a vast amount of knowledge, while updating knowledge in teaching is limited, along with the unchanged time in the classroom. Therefore, integrating bioinformatics into teaching will be effective in teaching biology today. However, the big challenge is that pedagogical university students have yet to learn the basic knowledge and skills of bioinformatics, so they have difficulty and confusion when using it. However, the big challenge is that pedagogical university students have yet to learn the basic knowledge and skills of bioinformatics, so they have difficulty and confusion when using it in biology teaching. This dataset includes survey results on high school teachers, teacher training curriculums and pedagogical students in Vietnam. The highlights of this dataset are six basic principles and four steps of bioinformatics integration in teaching biology at high schools, with illustrative examples. The principles and approaches of integrating Bioinformatics into biology teaching improve the quality of biology teaching and promote STEM education in Vietnam and developing countries.

This dataset represents an example for a textbook: Genomics and Proteomics for Clinical Discovery and Development, http://www.springer.com/life+sciences/systems+biology+and+bioinformatics/book/978-94-017-9201-1 ReAnalysis of the work Proteomic analysis of apical microvillous membranes of syncytiotrophoblast cells reveals a high degree of similarity with lipid rafts. Experiment described at: Paradela et al. J Proteome Res. 2005 Nov-Dec;4(6):2435-41 PMID: 16335998. For protein annotations, revise, Medina-Aunon et at. Proteomics. 2010 Sep;10(18):3262-71 PMID: 20707001. Genomics and Proteomics for Clinical Discovery and Development Translational Bioinformatics Volume 6, 2014, pp 41-68

Comprehensive example single-cell RNA sequencing dataset with marker genes specifically designed for testing and demonstrating AI-powered cell type annotation capabilities. This dataset includes representative cell clusters with known markers for validation purposes.