Twitter notes of SACMDA3

Several weeks ago, the 3rd Workshop on Statistical and Algorithmic Challenges in Microbiome Data Analysis was hosted at the Simons Foundation. I did not attend, but with the magic of Twitter and the internet, I could follow along with the generous efforts of live-tweeting with the hashtag #SACMDA3.

Here are my notes on what happened.

First day

A simple but crucial question in the world of microbiome science.

Incredibly impressed by the morning talks at #SACMDA3#SACMDA. “What’s in my sample?” is still an open and important question — and it’s to hear from both new and senior researchers working on strain id & tax. assignment qtns! Great work, Nidhi, Chris, Rachel & Francesco!!

— Amy Willis (@AmyDWillis) April 1, 2019

Great thread, summarizing the first day of the workshop, along with links to papers, GitHub and Bitbucket code repositories, and comments.

Excited to be at #SACMDA3, third annual-ish Workshop on Statistical and Algorithmic Challenges in Microbiome Data Analysis, sponsored by @MITMicrobiome and @SimonsFdn @FlatironCCB https://t.co/Gy4owbvZtK

— Claire Duvallet (@cduvallet) April 1, 2019

A reminder of our role as computationalists.

Another lol moment (for me) here at #SACMDA3: "Goal: narrow down the search space for experimentalists." #BenevolentComputationalists

— Claire Duvallet (@cduvallet) April 1, 2019

“A big challenge in computational metagenomics: Lacking consensus about benchmarking datasets, evaluation procedures, and metrics complicates prior performance assessments.”

Echo that! #SACMDA3 pic.twitter.com/7vHXUmeunc

— Erika Ganda (@EKGanda) April 1, 2019

Second day

Another great thread by Clarie Duvallet for the second day.

Day 2 of #SACMDA3 kicked off by @KnightLabNews emphasizing the need to get message about appropriate statistical techniques for data analysis to the entire #microbiome community. Ends with: anyone here wanna write a review paper with him, @ejalm, and @RichBonneauNYU? 😉

— Claire Duvallet (@cduvallet) April 2, 2019

This blog post on Statistical Methods for Microbiome Data was mentioned, especially on the section for a novel method for contamination removal.

“Every metagenomic measurement is wrong.”

This public service announcement brought to you by @bejcal and @mikemc423 #SACMDA3 pic.twitter.com/Sw1nNyuKIk

— Amy Willis (@AmyDWillis) April 2, 2019

Challenges when working with Longitudinal Multidomain data

• Data Quality: Heterogeneity, unwanted sourcecs of variation.
• Building modles from the data.
• Interpretation of analytic output.
• Multiple (dual) dependencies
• Multidomain, need for registration.
• Uncertainty quantification and inference.
• Reproducibility of results across labs, experimental conditions and users.

Susan Holmes talking about the many challenges when working with data. #SACMDA3 pic.twitter.com/KaWcw9gDOV

— Erika Ganda (@EKGanda) April 2, 2019

Software packages mentioned

• CONCOCT
  • “Unsupervised binning of metagenomic contigs by using nucleotide composition, coverage data in multiple samples and linkage data from paired end reads.”
  • Paper
  • GitHub
• Pandora
  • “Pan-genome inference with long error-prone sequencing reads”
  • In active development
  • GitHub
• DESMAN
  • “DESMAN identifies variants in core genes and uses co-occurrence across samples to link variants into haplotypes and abundance profiles.”
  • This tool was not mentioned directly, but appears to be the predecessor to a new tool that was mentioned
  • Paper
• PhyloPhlAn (Version 2)
  • “PhyloPhlAn is a computational pipeline for reconstructing highly accurate and resolved phylogenetic trees based on whole-genome sequence information.”
  • Paper
  • Bitbucket
• FEAST
  • “Fast expectation maximization source tracking”
  • GitHub
• rhapsody
  • “Neural networks for microbe-metabolite interaction analysis”
  • GitHub
• mixOmics
  • “mixOmics: An R package for ‘omics feature selection and multiple data integration”
  • Website
  • Paper

Papers

• Pasolli et al., “Extensive Unexplored Human Microbiome Diversity Revealed by Over 150,000 Genomes from Metagenomes Spanning Age, Geography, and Lifestyle”. Cell (2019)
• Wirbel et al., “Meta-analysis of fecal metagenomes reveals global microbial signatures that are specific for colorectal cancer”. Nature Medicine (2019)
• Sczyrba et al.,”Critical Assessment of Metagenome Interpretation—a benchmark of metagenomics software”. Nature Methods (2017)
• McLaren, Willis, Callahan. “Consistent and correctable bias in metagenomic sequencing measurements” bioRxiv (2019)
• Fernandes et al., “A reproducible effect size is more useful than an irreproducible hypothesis test to analyze high throughput sequencing datasets” arXiv (2018)
• Washburne et al., “Phylogenetic factorization of compositional data yields lineage-level associations in microbiome datasets”. PeerJ (2017)
• Washburne et al., “Phylofactorization: a graph-partitioning algorithm to identify phylogenetic scales of ecological data”. bioRxiv
• Rivera-Pinto et al., “Balances: a New Perspective for Microbiome Analysis”
• Silverman et al., “Bayesian Multinomial Logistic Normal Models through Marginally Latent Matrix-T Processes”. arXiv (2019)

Reflection

There was a lot of concepts, papers, and code repositories I couldn’t cover. Feel free to browse the Twitter feed for more.

Things to share

Sorting lines in Vim

You can sort lines in Vim with :{range}sort u. The u is to remove duplicate lines. More can be found at :help sort.

Spelling files created by Vim

For UTF-8 files, Vim creates two files:

• en.utf-8.add
• en.utf-8.add.spl

One discussion about them suggest one is plain text and the other is the corresponding compiled binary. I was editing my spelling files and was curiou about the difference between them.

The en.utf-8.add file is plain text and I have version controlled it for consistence across my computers.

The compiled .spl file can be recreated using

:mkspell ~/.vim/spell/en.uft-8.add