Tesorai FAQs

Tesorai Search currently provides core elements as output: PSM and peptide identification tables, as well as a report with a few data diagnostic plots. Here are a few features that we’ll be adding soon: protein inference, quantification, and support for TMT data and additional post-translational modification. Let us know what you would like to see added next: feedback@tesorai.com.

We use Google Cloud Storage (GCS) to store your data, which is designed with multiple layers of protection. Here are several key measures we take to safeguard your information:

• Robust Encryption: All data stored on GCS is encrypted at rest and in transit. This means your data is encrypted before it is written to disk and is also protected as it travels between our servers and GCS, preventing unauthorized access.
  ‍
• Compliance and Certifications: GCS complies with major security standards and certifications such as ISO 27001, which helps protect data from security threats or breaches. By complying with these standards, we ensure that our security measures meet international security guidelines.
  ‍
• Access Controls: We implement strict access controls and auditing capabilities. Access to data is tightly controlled and limited to authorized personnel only, based on their role. We also use advanced identity and access management policies to prevent unauthorized access.
  ‍
• Regular Security Audits: We regularly review and update our security practices. This includes conducting security audits and vulnerability scans to ensure that our defenses remain effective against new threats.
  ‍
• Data Redundancy: GCS provides high durability via data redundancy. This means your data is automatically replicated across multiple locations to prevent data loss due to hardware failures or other incidents.

By leveraging Google Cloud's robust infrastructure and adhering to best practices, we ensure that your data is secure and protected at all times. If you have any more questions about our data security measures, please feel free to contact us.

Pricing structures are based on account types.

‍

Academic/research: users from academic or research institutions have access to the Tesorai platform for FREE, users can process up to 500 RAW files per month.

‍

Commercial: Users from commercial organizations can reach out directly to info@tesorai.com for enterprise pricing. Commercial Users are welcome to process up to 20 RAW initially. Reach out to info@tesorai.com and we can find the most cost effective plan for your needs.

‍

By choosing Tesorai Search, you are investing in a robust, cloud-based solution that eliminates hidden costs associated with workstation software, such as extensive time for installation and maintenance, as well as longer wait times for data processing. Our platform is designed to save you time and effort, enabling you to focus more on your research and less on the logistical aspects of data processing.

No, we do not share your data with third parties. Our commitment to your privacy and data security is paramount. Here is how we handle your data:

• Strict Privacy Policy: We adhere to a strict privacy policy that prohibits the sharing of your data with any external companies or third parties. This ensures that your personal and business data remains confidential and is used solely to enhance your experience with our service.
  ‍
• Internal Use Only: Any data collected from you is used exclusively for internal purposes, such as improving our services, providing support, and making informed decisions that benefit our user community. We do not sell, rent, or trade your information with any external entities.
  ‍
• We value your trust and are committed to protecting your data with the highest standards of privacy and security. If you have any further questions or concerns about how we handle your data, please feel free to reach out to our support team.

The current model version of Tesorai Search can support DDA datasets. The team is currently evaluating models that can also support TMT and DIA datasets. Provide your email address here [scroll to bottom of the page to subscribe to our newsletter] to be notified for when an early launch of those models are ready to be tested. Our goal is to be instrument agnostic. We tested our algorithm on TimsTOF data, and found similar or better results than the algorithms used to search the TOF data (i.e., MSFragger). An interesting note is that the model has not yet been trained or optimized on TimsTOF data and yet still works fairly well. We will be releasing an optimized (Tims)TOF and TMT model soon. Sign up for updates on pending release/soft launch dates.

The processing time for a dataset in Tesorai Search depends on several factors, including the raw file sizes, the number of spectra in the raw files, the number of proteins in the fasta file, and the digestion enzyme used. However, Tesorai Search leverages the power of cloud computing, which allows us to dynamically allocate resources based on the demand. This means that as your dataset grows or as more complex queries are executed, our system can scale the processing power to match, ensuring consistent performance.

‍

Estimation Guidelines (using standard fasta files):

‍

Small datasets (up to 10 raw files): 40 minutes
Medium datasets (10 to 100 raw files): 45 minutes
Large datasets (over 100 raw files): 1 hour (for example, we recently processed 500 raw files in under 50 minutes)

‍

We understand the importance of efficiency in data processing and continuously work to improve the performance of Tesorai Search. Should you have specific requirements or need more detailed information, our technical support team is available to provide assistance and discuss how we can best meet your needs.

We currently limit the fasta file size to 100 MB. This is temporary as we continue to expand our computational framework. If you have a larger file or specific requirements that exceed this limitation, we encourage you to reach out to us directly.

Tesorai Search does not limit the number of files you can upload and process. You have the flexibility to upload and process as many files as you need. However, there is a size limit of 5 GB per individual file; note that this is temporary as we continue to expand our computational framework. If you have larger files or specific requirements that exceed this limitation, we encourage you to reach out to us directly.

Software updates will be automatic; we’ll be adding features continuously. We welcome your feedback as we release new versions, and suggestions for new features that would further improve your experience.

The time it takes to upload data to Tesorai Search generally depends on a few key factors, including the size of the files you are uploading and the current internet conditions. While the actual upload speed is influenced by your internet bandwidth, our platform is optimized to handle uploads as efficiently as possible. For example, for a dataset with 5 files and 1 Gb in size each, and on a typical 50 Mbps upload internet speed, the data upload would only take a couple of minutes.

No need for installation; Tesorai Search is web-based, unlike many other tools that require installation and frequent updates on workstations. If your organization requires an on-prem solution, please reach out to info@tesorai.com.

Yes – the current version can help identification of protein modifications, as long as they’re included in the search fasta file. The standard, sample-preparation modifications such as cysteine carbamylation, methionine oxidation, deamidation, and acetylation are also supported by default. Other PTMs, such as phosphorylation, ubiquitination, glycosylation and more are being added as we speak and will become available soon. White paper coming soon.

Onboarding and training with Tesorai Search is designed to be straightforward and quick, ensuring you can start leveraging our tool with minimal delay. Unlike other programs that can be difficult to install and configure, Tesorai Search is a cloud-based platform with a user-friendly interface featuring simple drag-and-drop functionality.

The algorithm behind Tesorai Search is trained and evaluated on a variety of data types, and works with different instrument manufacturers (note: the beta version only supports upload of Thermo raw files, but options to upload raw files from other MS vendors will be available soon. Tesorai Search also works with several digestion enzymes and unspecific digestion (i.e., immunopeptidomics), collision energies, and fragmentation types. The model performs well on standard, single-species datasets as well as more complex meta-proteomics studies.

Tesorai Search and Tesorai Refine can both advance research efforts in pre-clinical drug development and protein characterization of drug development, including target discovery, toxicity prediction, biomarker discovery, etc. We also work with biotech/pharma partners with customized models and service engagements – contact us to learn more. Moreover, our tool is not limited to human samples, and we’d welcome use with other species and applications beyond healthcare and human biology, including for agriculture or wastewater analysis.

Mass spectrometers collect a massive amount of spectral data but the majority of those mass spectra remain unidentified or unannotated - current algorithms can fail to identify up to 80% of tandem spectra. Tesorai Search leverages the most recent AI advances to enable users to increase identification of peptides and analytes in their samples despite these challenges. Other tools, such as Percolator, have also attempted to leverage other machine learning approaches for rescoring, boosting peptide and protein identification rates. But the results from recent tools that rely on boosting rates with the use of Percolator come at a cost, as they are trained to separate the targets from the decoys, which can lead to inaccurate control of the false-discovery rate [Freestone et al].

‍

In contrast, Tesorai Search increases peptide identification without needing Percolator. Our advanced approach to peptide-spectrum matching is based on a pre-trained large deep learning model which does not utilize decoys during training and does not require training a new model for every new sample. We trained the Tesorai model on over 100M real peptide-spectrum pairs and demonstrate that the approach performs robustly across a wide range of use-cases including standard trypsin-digested human samples, immunopeptidomics, and metaproteomics, single-cell and isobaric-labeled samples. In addition to providing robust FDR control, our method increases identifications by 20% or more compared to other advanced methods.