Release Notes

For people amusement, while reviewing new studies on PubMed, I got a pleasant surprise! My citizen science project (MicrobiomePrescription) is referenced in BMC Bioinformatics article published June 3 2021. Adding to my prior mentions in New Scientist and The Lancet. “Dysbiosis of the gut microbiome is a risk factor for osteoarthritis in older female adults: a case control study

Most errors will automatically send emails about the issue and they should be resolved within 24 hrs. If something is still broken after 24 hours, please email Research@MicrobiomePrescription.com.

Currently the online database consists of 50 million rows of data.

Caveat Emptor 

This is a “best effort” project. Data may be incorrectly interpreted or entered on occasion. It is believed that the entry accuracy is between 95% and 98%. Links to the source study are provided for people to verify information before relying on it. There is no claim of “correctness”, just reasonable efforts.

2024

  • March 4-18: Down with COVID — bugs may occur. Please report as you find them
  • March 13: User identified processing issues with data from Extensive impact of non-antibiotic drugs on human gut bacteria. All old data (over 100,000 facts) deleted and data re-entered.
    • The relative impact of each was incorporated in the revised fact.
    • Some new drugs, substances were added, for example Piracetam, the nootropic.
    • Some suggestions are expected to shift.
  • March 12: Add Mold Index to site, fixed Symptom forecasts to be out of 100.
  • March 11: Restored Adding Symptoms functionality. Broke by a refactoring elsewhere
  • Match 10: Refactor Common Measure to use lab specific data for percentile, updated the ? Meaning for missing items
  • March 9 Taxa Association to symptoms expanded to allow filtering by different taxa ranks (not just genus). For each option, a priori suggestions were added.
  • March 2 – Monthly update of Percentile and other regressions scheduled to occur on the 2nd of each month, Updates will usually cause only minor shifts of suggestions unless rare bacteria is involved, then more dramatic changes may occur.
  • Feb 28 – Refactor Genus Associations for faster execution and expanded to include uBiome and Thorne in the regression.
  • Feb 26 – Refactor PubMed Medical Conditions providing more data for evaluation and a separate page/panel
  • Feb 22 – Deleted a citation (on Cadium) and it’s related facts because the study link was very wrong and the correct source could not identified. One of the backup links went to a different article, that is now annotated Retracted article. Some changes of suggestions could occur as a result.
  • Feb 21 – Audited Taxa Percentile and found that 1.5% were computed too low. This occurred on rare taxa.
  • Feb 16 – Added Absolutely safe suggestions filter to consensus view
  • Feb 14 – Added and tested two ways of selecting bacteria based on symptoms
    Video on above 2 features: https://www.youtube.com/watch?v=CdrRCyQNzfs
  • Feb 9th, User pointed out an error in a few Bacteria description (wrong bacteria described). Implemented audit and correction of 1,336 descriptions.
  • Feb 7th, deprecated old symptom forecasting method, replacing with new genus based algorithm.
  • Feb 6th, Investigated issue with two uploads of the same data reporting different results. Explanation:
    When an upload happens, I create missing hierarchical levels by summing up the children. This means that two identical samples uploaded at different times may different missing levels if the Hierarchy table was updated or modified between samples.
  • Feb 6th, Refactor the PubMed studies to support selecting multiple conditions concurrently. 16% is a toss up, below 16% is unlikely to have a condition. The higher the percentage match, the more likely.
  • Feb 4 — Finished refactoring of Percentiles on Enzymes, Compound and Taxons. The refactor reduced the database size by over 50% and should eliminated most stale data. There may be some bugs as side-effect. Please report them ASAP.
  • Feb 3 Remove this obsolete section. It was an early punt prior to getting consensus working well. It used only one method and as a result, it was often in disagreement with the consensus.
  • Feb 2 Revised KEGG section to allow individual suggestions derived from trying to tune using Kegg Compounds. See Technical Demo: Using Compounds and Enzymes to get Suggestions (youtube.com)
  • Feb 1 – Satisfied Jan 25th bug has been fixed
  • Jan 30 — Removed the old metabolite section. It had been marked obsolete for over a year and it was time to remove it after confirming that it was likely giving incomplete information. See Technical Notes: Balancing Priorities for Fixing the Gut
  • Jan 26, removed quasi duplicate citations. For some studies both the parent and the child are mentioned as shifting. We infer parent and descendants using each bacteria, this resulted in TWO records for some bacteria — the direct citation and a second citation thru a child bacteria. We identifies some 13,000 such records (in over 2.5 million) and removed these secondary records.
  • Jan 25, Significant Bug found. It may have some impact on suggestions (bug impacted just 1 of 4 methods) At one spot, I have percentiles  as 0-100 in the sample; but in reference values I used 0.0-1.0. This impacts KM only. Suggestions may change slightly
  • Jan 25, Added Taxa Rank to Bacteria Details page
  • Jan 24, Added Citizen Science symptoms to Bacteria Details page
  • Jan 20, Added move lab ranges to database.
  • Jan 17, Added TeleTest as a supported lab for transcription.
  • Jan 11, fixed issues of citations link not working on Consensus View
  • Jan 9th, fixed issue on https://microbiomeprescription.com/Sample/DoubleSignificant when there is a long list of parameters
  • Fixed issue with Update button and then refactor so percentiles are updated with live data (ignoring duplicate uploads). Recent refactors allow that to be more efficiently done in a reasonable length of time.
    • Taxon
    • Enzymes
    • Compound
  • Complete Update on ALL uploaded samples is done once a week.