70 avsnitt • Längd: 65 min • Oregelbundet
A podcast about computational biology, bioinformatics, and next generation sequencing.
The podcast the bioinformatics chat is created by Roman Cheplyaka. The podcast and the artwork on this page are embedded on this page using the public podcast feed (RSS).
In this episode, Marie Sadler talks about her recent Cell Genomics paper, Multi-layered genetic approaches to identify approved drug targets.
Previous studies have found that the drugs that target a gene linked to the disease are more likely to be approved. Yet there are many ways to define what it means for a gene to be linked to the disease. Perhaps the most straightforward approach is to rely on the genome-wide association studies (GWAS) data, but that data can also be integrated with quantitative trait loci (eQTL or pQTL) information to establish less obvious links between genetic variants (which often lie outside of genes) and genes. Finally, there’s exome sequencing, which, unlike GWAS, captures rare genetic variants. So in this paper, Marie and her colleagues set out to benchmark these different methods against one another.
Listen to the episode to find out how these methods work, which ones work better, and how network propagation can improve the prediction accuracy.
Links:
Thank you to Jake Yeung, Michael Weinstein, and other Patreon members for supporting this episode.
Today on the podcast we have Tomasz Kociumaka and Dominik Kempa, the authors of the preprint Collapsing the Hierarchy of Compressed Data Structures: Suffix Arrays in Optimal Compressed Space.
The suffix array is one of the foundational data structures in bioinformatics, serving as an index that allows fast substring searches in a large text. However, in its raw form, the suffix array occupies the space proportional to (and several times larger than) the original text.
In their paper, Tomasz and Dominik construct a new index, δ-SA, which on the one hand can be used in the same way (answer the same queries) as the suffix array and the inverse suffix array, and on the other hand, occupies the space roughly proportional to the gzip’ed text (or, more precisely, to the measure δ that they define — hence the name).
Moreover, they mathematically prove that this index is optimal, in the sense that any index that supports these queries — or even much weaker queries, such as simply accessing the i-th character of the text — cannot be significantly smaller (as a function of δ) than δ-SA.
Links:
Thank you to Jake Yeung and other Patreon members for supporting this episode.
In this episode, David Dylus talks about Read2Tree, a tool that builds alignment matrices and phylogenetic trees from raw sequencing reads. By leveraging the database of orthologous genes called OMA, Read2Tree bypasses traditional, time-consuming steps such as genome assembly, annotation and all-versus-all sequence comparisons.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
This is the third and final episode in the AlphaFold series, originally recorded on February 23, 2022, with Amelie Stein, now an associate professor at the University of Copenhagen.
In the episode, Amelie explains what 𝛥𝛥G is, how it informs us whether a particular protein mutation affects its stability, and how AlphaFold 2 helps in this analysis.
A note from Amelie:
Something that has happened in the meantime is the publication of methods that predict 𝛥𝛥G with ML methods, so much faster than Rosetta. One of them, RaSP, is from our group, while ddMut is from another subset of authors of the AF2 community assessment paper.
Other links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
This is the second episode in the AlphaFold series, originally recorded on February 14, 2022, with Janani Durairaj, a postdoctoral researcher at the University of Basel.
Janani talks about how she used shape-mers and topic modelling to discover classes of proteins assembled by AlphaFold 2 that were absent from the Protein Data Bank (PDB).
The bioinformatics discussion starts at 03:35.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, originally recorded on February 9, 2022, Roman talks to Pedro Beltrao about AlphaFold, the software developed by DeepMind that predicts a protein’s 3D structure from its amino acid sequence.
Pedro is an associate professor at ETH Zurich and the coordinator of the structural biology community assessment of AlphaFold2 applications project, which involved over 30 scientists from different institutions.
Pedro talks about the origins of the project, its main findings, the importance of the confidence metric that AlphaFold assigns to its predictions, and Pedro’s own area of interest — predicting pockets in proteins and protein-protein interactions.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Jacob Schreiber interviews Žiga Avsec about a recently released model, Enformer. Their discussion begins with life differences between academia and industry, specifically about how research is conducted in the two settings. Then, they discuss the Enformer model, how it builds on previous work, and the potential that models like it have for genomics research in the future. Finally, they have a high-level discussion on the state of modern deep learning libraries and which ones they use in their day-to-day developing.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
The Bioinformatics Contest is back this year, and we are back to discuss it!
This year’s contest winners Maksym Kovalchuk (1st prize) and Matt Holt (2nd prize) talk about how they approach participating in the contest and what strategies have earned them the top scores.
Timestamps and links for the individual problems:
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Apostolos Chalkis presents sampling steady states of metabolic networks as an alternative to the widely used flux balance analysis (FBA). We also discuss dingo, a Python package written by Apostolos that employs geometric random walks to sample steady states. You can see dingo in action here.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Jacob Schreiber interviews Da-Inn Erika Lee about data and computational methods for making sense of 3D genome structure. They begin their discussion by talking about 3D genome structure at a high level and the challenges in working with such data. Then, they discuss a method recently developed by Erika, named GRiNCH, that mines this data to identify spans of the genome that cluster together in 3D space and potentially help control gene regulation.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Michael Love joins us to talk about the differential gene expression analysis from bulk RNA-Seq data.
We talk about the history of Mike’s own differential expression package, DESeq2, as well as other packages in this space, like edgeR and limma, and the theory they are based upon. Mike also shares his experience of being the author and maintainer of a popular bioninformatics package.
Links:
And a more comprehensive set of links from Mike himself:
limma, the original paper and limma-voom:
https://pubmed.ncbi.nlm.nih.gov/16646809/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4053721/
edgeR papers:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2796818/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3378882/
The recent manuscript mentioned from the Kendziorski lab, which has a Gamma-Poisson hierarchical structure, although it does not in general reduce to the Negative Binomial:
https://doi.org/10.1101/2020.10.28.359901
We talk about robust steps for estimating the middle of the dispersion prior distribution, references are Anders and Huber 2010 (DESeq), Eling et al 2018 (one of the BASiCS papers), and Phipson et al 2016:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3218662/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6167088/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5373812/
The Stan software:
https://mc-stan.org/
We talk about using publicly available data as a prior, references I mention are the McCall et al paper using publicly available data to ask if a gene is expressed, and a new manuscript from my lab that compares splicing in a sample to GTEx as a reference panel:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3013751/ https://doi.org/10.1101/856401
Regarding estimating the width of the dispersion prior, references are the Robinson and Smyth 2007 paper, McCarthy et al 2012 (edgeR), and Wu et al 2013 (DSS):
https://pubmed.ncbi.nlm.nih.gov/17881408/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3378882/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3590927/
Schurch et al 2016, a RNA-seq dataset with many replicates, helpful for benchmarking:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4878611/
Stephens paper on the false sign rate (ash):
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5379932/
Heavy-tailed distributions for effect sizes, Zhu et al 2018:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6581436/
I credit Kevin Blighe and Alexander Toenges, who help to answer lots of DESeq2 questions on the support site:
https://www.biostars.org/u/41557/
https://www.biostars.org/u/25721/
The EOSS award, which has funded vizWithSCE by Kwame Forbes, and nullranges by Wancen Mu and Eric Davis:
https://chanzuckerberg.com/eoss/proposals/ensuring-reproducible-transcriptomic-analysis-with-deseq2-and-tximeta/
https://kwameforbes.github.io/vizWithSCE/
https://nullranges.github.io/nullranges/
One of the recent papers from my lab, MRLocus for eQTL and GWAS integration:
https://mikelove.github.io/mrlocus/
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Lindsay Pino discusses the challenges of making quantitative measurements in the field of proteomics. Specifically, she discusses the difficulties of comparing measurements across different samples, potentially acquired in different labs, as well as a method she has developed recently for calibrating these measurements without the need for expensive reagents. The discussion then turns more broadly to questions in genomics that can potentially be addressed using proteomic measurements.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, we learn about B cell maturation and class switching from Hamish King. Hamish recently published a paper on this subject in Science Immunology, where he and his coauthors analyzed gene expression and antibody repertoire data from human tonsils. In the episode Hamish talks about some of the interesting B cell states he uncovered and shares his thoughts on questions such as «When does a B cell decide to class-switch?» and «Why is the antibody isotype correlated with its affinity?»
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Jacob Schreiber interviews Molly Gasperini about enhancer elements. They begin their discussion by talking about Octant Bio, and then dive into the surprisingly difficult task of defining enhancers and determining the mechanisms that enable them to regulate gene expression.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Polygenic risk scores (PRS) rely on the genome-wide association studies (GWAS) to predict the phenotype based on the genotype. However, the prediction accuracy suffers when GWAS from one population are used to calculate PRS within a different population, which is a problem because the majority of the GWAS are done on cohorts of European ancestry.
In this episode, Bárbara Bitarello helps us understand how PRS work and why they don’t transfer well across populations.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, we chat about phylogenetics with Xiang Ji. We start with a general introduction to the field and then go deeper into the likelihood-based methods (maximum likelihood and Bayesian inference). In particular, we talk about the different ways to calculate the likelihood gradient, including a linear-time exact gradient algorithm recently published by Xiang and his colleagues.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Markus Schmidt explains how seeding in read alignment works. We define and compare k-mers, minimizers, MEMs, SMEMs, and maximal spanning seeds. Markus also presents his recent work on computing variable-sized seeds (MEMs, SMEMs, and maximal spanning seeds) from fixed-sized seeds (k-mers and minimizers) and his Modular Aligner.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Jacob Schreiber interviews Devin Schweppe about the analysis of mass spectrometry data in the field of proteomics. They begin by delving into the different types of mass spectrometry methods, including MS1, MS2, and, MS3, and the reasons for using each. They then discuss a recent paper from Devin, Full-Featured, Real-Time Database Searching Platform Enables Fast and Accurate Multiplexed Quantitative Proteomics that involved building a real-time system for quantifying proteomic samples from MS3, and the types of analyses that this system allows one to do.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Will Freyman talks about identity-by-descent (IBD): how it’s used at 23andMe, and how the templated positional Burrows-Wheeler transform can find IBD segments in the presence of genotyping and phasing errors.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Jacob Schreiber interviews David Kelley about machine learning models that can yield insight into the consequences of mutations on the genome. They begin their discussion by talking about Calico Labs, and then delve into a series of papers that David has written about using models, named Basset and Basenji, that connect genome sequence to functional activity and so can be used to quantify the effect of any mutation.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Jacob Schreiber interviews Jill Moore about recent research from the ENCODE Project. They begin their discussion with an overview and goals of the ENCODE Project, and then discuss a bundle of papers that were recently published in various Nature journals and the flagship paper, Expanded encyclopaedias of DNA elements in the human and mouse genomes. They conclude their discussion by talking about the challenges with managing a large project as a trainee in a consortium setting.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In systems biology, Boolean networks are a way to model interactions such as gene regulation or cell signaling. The standard interpretations of Boolean networks are the synchronous, asynchronous, and fully asynchronous semantics.
In this episode, Loïc Paulevé explains how the same Boolean networks can be interpreted in a new, “most permissive” way. Loïc proved mathematically that his semantics can reproduce all behaviors achievable by a compatible quantitative model, whereas the traditional interpretations in general cannot. Furthermore, it turns out that deciding whether a certain state in a Boolean network is reachable can be done much more efficiently in MPBNs than in the traditional interpretations.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Jacob Schreiber interviews Marinka Zitnik about applications of machine learning to drug development. They begin their discussion with an overview of open research questions in the field, including limiting the search space of high-throughput testing methods, designing drugs entirely from scratch, predicting ways that existing drugs can be repurposed, and identifying likely side-effects of combining existing drugs in novel ways. Focusing on the last of these areas, they then discuss one of Marinka’s recent papers, Modeling polypharmacy side effects with graph convolutional networks.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
NGLess is a programming language specifically targeted at next generation sequencing (NGS) data processing. In this episode we chat with its main developer, Luis Pedro Coelho, about the benefits of domain-specific languages, pros and cons of Haskell in bioinformatics, reproducibility, and of course NGLess itself.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, I continue to talk (but mostly listen) to Sergey Koren and Sergey Nurk. If you missed the previous episode, you should probably start there. Otherwise, join us to learn about HiFi reads, the tradeoff between read length and quality, and what tricks HiCanu employs to resolve highly similar repeats.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Sergey Nurk and Sergey Koren from the NIH share their thoughts on genome assembly. The two Sergeys tell the stories behind their amazing careers as well as behind some of the best known genome assemblers: Celera assembler, Canu, and SPAdes.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Porcupine is a molecular tagging system—a way to tag physical objects with pieces of DNA called molecular bits, or molbits for short. These DNA tags then can be rapidly sequenced on an Oxford Nanopore MinION device without any need for library preparation.
In this episode, Katie Doroschak explains how Porcupine works—how molbits are designed and prepared, and how they are directly recognized by the software without an intermediate basecalling step.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Will Townes proposes a new, simpler way to analyze scRNA-seq data with unique molecular identifiers (UMIs). Observing that such data is not zero-inflated, Will has designed a PCA-like procedure inspired by generalized linear models (GLMs) that, unlike the standard PCA, takes into account statistical properties of the data and avoids spurious correlations (such as one or more of the top principal components being correlated with the number of non-zero gene counts).
Also check out Will’s paper for a feature selection algorithm based on deviance, which we didn’t get a chance to discuss on the podcast.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, we hear from Amatur Rahman and Karel Břinda, who independently of one another released preprints on the same concept, called simplitigs or spectrum-preserving string sets. Simplitigs offer a way to efficiently store and query large sets of k-mers—or, equivalently, large de Bruijn graphs.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Kris Parag is here to teach us about the mathematical modeling of infectious disease epidemics. We discuss the SIR model, the renewal models, and how insights from information theory can help us predict where an epidemic is going.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Does a given bacterial gene live on a plasmid or the chromosome? What other genes live on the same plasmid?
In this episode, we hear from Sergio Arredondo-Alonso and Anita Schürch, whose projects mlplasmids and gplas answer these types of questions.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Benjamin Callahan talks about some of the issues faced by microbiologists when conducting amplicon sequencing and metagenomic studies. The two main themes are:
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Luke Anderson-Trocmé talks about his findings from the 1000 Genomes Project. Namely, the early sequenced genomes sometimes contain specific mutational signatures that haven’t been replicated from other sources and can be found via their association with lower base quality scores. Listen to Luke telling the story of how he stumbled upon and investigated these fake variants and what their impact is.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, I talk with Irineo Cabreros about causality. We discuss why causality matters, what does and does not imply causality, and two different mathematical formalizations of causality: potential outcomes and directed acyclic graphs (DAGs). Causal models are usually considered external to and separate from statistical models, whereas Irineo’s new paper shows how causality can be viewed as a relationship between particularly chosen random variables (potential outcomes).
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, we hear from Romain Lopez and Gabriel Misrachi about scVI—Single-cell Variational Inference. scVI is a probabilistic model for single-cell gene expression data that combines a hierarchical Bayesian model with deep neural networks encoding the conditional distributions. scVI scales to over one million cells and can be used for scRNA-seq normalization and batch effect removal, dimensionality reduction, visualization, and differential expression. We also discuss the recently implemented in scVI automatic hyperparameter selection via Bayesian optimization.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Even though the double-stranded DNA has the famous regular helical shape, there are small variations in the geometry of the helix depending on what exact nucleotides its made of at that position.
In this episode of the bioinformatics chat, Hassan Samee talks about the role the DNA shape plays in recognition of the DNA by DNA-binding proteins, such as transcription factors. Hassan also explains how his algorithm, ShapeMF, can deduce the DNA shape motifs from the ChIP-seq data.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
An αβ T-cell receptor is composed of two highly variable protein chains, the α chain and the β chain. However, based only on bulk DNA or RNA sequencing it is impossible to determine which of the α chain and β chain sequences were paired in the same receptor.
In this episode, Kristina Grigaityte talks about her analysis of 200,000 paired αβ sequences, which have been obtained by targeted single-cell RNA sequencing. Kristina used the power law distribution to model the T-cell clone sizes, which led her to reject the commonly held assumptions about the independence of the α and β chains. We also talk about Bayesian inference of power law distributions and about mixtures of power laws.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Modern genome assembly projects are often based on long reads in an attempt to bridge longer repeats. However, due to the higher error rate of the current long read sequencers, assemblers based on de Bruijn graphs do not work well in this setting, and the approaches that do work are slower.
In this episode, Mikhail Kolmogorov from Pavel Pevzner’s lab joins us to talk about some of the ideas developed in the lab that made it possible to build a de Bruijn-like assembly graph from noisy reads. These ideas are now implemented in the Flye assembler, which performs much faster than the existing long read assemblers without sacrificing the quality of the assembly.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, we hear from Jacob Schreiber about his algorithm, Avocado.
Avocado uses deep tensor factorization to break a three-dimensional tensor of epigenomic data into three orthogonal dimensions corresponding to cell types, assay types, and genomic loci. Avocado can extract a low-dimensional, information-rich latent representation from the wealth of experimental data from projects like the Roadmap Epigenomics Consortium and ENCODE. This representation allows you to impute genome-wide epigenomics experiments that have not yet been performed.
Jacob also talks about a pitfall he discovered when trying to predict gene expression from a mix of genomic and epigenomic data. As you increase the complexity of a machine learning model, its performance may be increasing for the wrong reason: instead of learning something biologically interesting, your model may simply be memorizing the average gene expression for that gene across your training cell types using the nucleotide sequence.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
The third Bioinformatics Contest took place in February 2019.
Alexey Sergushichev, one of the organizers of the contest, and Gennady Korotkevich, the 1st prize winner, join me to discuss this year’s problems.
Timestamps and links for the individual problems:
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Hi-C is a sequencing-based assay that provides information about the 3-dimensional organization of the genome. In this episode, Simeon Carstens explains how he applied the Inferential Structure Determination (ISD) framework to build a 3D model of chromatin and fit that model to Hi-C data using Hamiltonian Monte Carlo and Gibbs sampling.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Long read sequencing technologies, such as Oxford Nanopore and PacBio, produce reads from thousands to a million base pairs in length, at the cost of the increased error rate. Trevor Pesout describes how he and his colleagues leverage long reads for simultaneous variant calling/genotyping and phasing. This is possible thanks to a clever use of a hidden Markov model, and two different algorithms based on this model are now implemented in the MarginPhase and WhatsHap tools.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
This time you’ll hear from Fabio Cunial on the topic of Markov models and space-efficient data structures. First we recall what a Markov model is and why variable-order Markov models are an improvement over the standard, fixed-order models. Next we discuss the various data structures and indexes that allowed Fabio and his collaborators to represent these models in a very small space while still keeping the queries efficient. Burrows-Wheeler transform, suffix trees and arrays, tries and suffix link trees, and more!
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, HoJoon Lee and Seung Woo Cho explain how to perform a CRISPR experiment and how to analyze its results. HoJoon and Seung Woo developed an algorithm that analyzes sequenced amplicons containing the CRISPR-induced double-strand break site and figures out what exactly happened there (e.g. a deletion, insertion, substitution etc.)
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Relief is a statistical method to perform feature selection. It could be used, for instance, to find genomic loci that correlate with a trait or genes whose expression correlate with a condition. Relief can also be made sensitive to interaction effects (known in genetics as epistasis).
In this episode, Trang Lê joins me to talk about Relief and her version of Relief called STIR (STatistical Inference Relief). While traditional Relief algorithms could only rank features and needed a user-supplied threshold to decide which features to select, Trang’s reformulation of Relief allowed her to compute p-values and make the selection process less arbitrary.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Kaushik Panda and Keith Slotkin come on the podcast to educate us about repetitive DNA and transposable elements. We talk LINEs, SINEs, LTRs, and even Sleeping Beauty transposons! Kaushik and Keith explain why repeats matter for your whole-genome analysis and answer listeners’ questions.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Antoine Limasset joins me to talk about NGS read correction. Antoine and his colleagues built the read correction tool Bcool based on the de Bruijn graph, and it corrects reads far better than any of the current methods like Bloocoo, Musket, and Lighter.
We discuss why and when read correction is needed, how Bcool works, and why it performs better but slower than k-mer spectrum methods.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, I talk to Fernando Portela, a software engineer and amateur scientist who works on RNA design — the problem of composing an RNA sequence that has a specific secondary structure.
We talk about how Fernando and others compete and collaborate in designing RNA molecules in the online game EteRNA and about Fernando’s new RNA design algorithm, NEMO, which outperforms all prior published methods by a wide margin.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode I’m joined by Chang Xu. Chang is a senior biostatistician at QIAGEN and an author of smCounter2, a low-frequency somatic variant caller. To distinguish rare somatic mutations from sequencing errors, smCounter2 relies on unique molecular identifiers, or UMIs, which help identify multiple reads resulting from the same physical DNA fragment.
Chang explains what UMIs are, why they are useful, and how smCounter2 and other tools in this space use UMIs to detect low-frequency variants.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Linear mixed models are used to analyze GWAS data and detect QTLs. Andrey Ziyatdinov recently released an R package, lme4qtl, that can be used to formulate and fit these models. In this episode, Andrey and I discuss linear mixed models, genome-wide association studies, and strengths and weaknesses of lme4qtl.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode Kristian Davidsen and Amrit Dhar present their project called SPURF. SPURF can predict the B cell receptor (BCR) substitution profile of a given clonal family based on a single representative sequence from that family. SPURF works by fitting a tensor regression model to publicly available Rep-seq data.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Gustavo Glusman explains his method of reducing a VCF file to a small “fingerprint”, which could be then used to detect duplicate genomes, infer relatedness, map the population structure, and more.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
The final round of Bioinformatics Contest 2018 was held on February 24-25th, and the qualification round took place two weeks earlier.
I invited the organizers of the contest, Alexey Sergushichev and Ekaterina Vyahhi, to discuss the problems and find out what it was like to organize the contest.
Timestamps for the problems:
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Amy Willis joins me to talk about good and bad ways to estimate taxonomic richness in microbial ecology studies.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Javier Quilez and I discuss what it’s like to be a bioinformatician, how to improve communication between the wet and dry labs and make the research more reproducible.
Make sure to read Javier’s paper we are discussing; it’s a light and entertaining read. The last author on this paper is Guillaume Filion, whom you may remember from the episode on generating functions.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Geoffrey Schiebinger explains how reconstructing developmental trajectories from single-cell RNA-seq data can be reduced to the mathematical problem called optimal transport.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Guillaume Filion recently published a preprint in which he applies generating functions, a concept from analytic combinatorics, to estimating the optimal seed length for read mapping.
In this episode, Guillaume and I attempt to explain the core concepts from analytic combinatorics and why they are useful in modeling sequences.
Links:
After we recorded the podcast, Guillaume wrote to me to clarify the relationship between read mapping and BLAST:
I looked into my notes about BLAST. The problem that it solves is the following: “Given that a local alignment has score S, what is the probability that it does not contain a word of score T or greater”? The background work of Karlin and Altschul is used to give a statistical significance for S (what is the probability that a “Smith-Waterman random walk” starting at height 0 would reach height S, i.e. what is the probability that aligning two random proteins would yield a score S). The authors write in the original paper “Theory does not yet exist to calculate the probability q that such segment pair will contain a word pair with a score of at least T. However, one argument suggests that q should depend exponentially upon the score of the MSP”.
This is the part that I did not remember well. MSP stands for Maximal Segment Pair, this is the “longest fragment” with “highest score” in the alignment. I thought that Karlin and Altschul solved this part as well, but the authors just go empirical and they calibrate the relationship between T and S with simulations.
I realize a little bit better now that my work is precisely about this problem that the authors of BLAST could not solve, but as you pointed out, I am attacking only a very specific sub-case that is much easier because the models of sequencing error are much simpler than protein evolution. BLAST is concerned with local alignment, so it wants to get all the hits with an MSP score above S. Short read mapping just wants the true location of the read, which does not really have the notion of a score S. But still, mathematically, it is equivalent to the case where S is a constant that depends only on the read size and the distribution of the score T depends only on the seed length and the error rate. I have a few ideas of how to use analytic combinatorics to solve the problem for proteins, but it is mostly complicated because the variable of interest T is a fractional numbers and not an integer…
So what is different from BLAST? The right answer (I think) is that BLAST finds all the hits with an MSP above statistical background, but it says nothing of the probability that the true location contains such an MSP, so it is hard to calibrate the heuristic for that specific problem. In reality, the parallel with BLAST is just the basic strategy: make a statistical model for your problem and use it to calibrate the heuristic.
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Jennifer Lu joins me to discuss species abundance estimation from metagenomic sequencing data.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Filippo Castiglione and I discuss different ways to model the immune system.
Links:
Special thanks to Martina Stoycheva for bringing this work to my attention.
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Linus Schumacher joins me to discuss mathematical models of collective cell migration and multidisciplinary research.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Valentine Svensson explains how he analyzes spatially-annotated single cell gene expression data using Gaussian processes.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
Michael Tessler and Christopher Mason join me to talk about their comparison of 16S amplicon sequencing and shotgun sequencing for quantifying microbial diversity.
Links:
Michael’s et al. 2016 paper that describes their original 16S study: A Global eDNA Comparison of Freshwater Bacterioplankton Assemblages Focusing on Large-River Floodplain Lakes of Brazil
The sequencing data for these studies is available from NCBI: PRJNA310230 (16S), PRJNA389803 (shotgun)
If you enjoyed this episode, please consider supporting the podcast on Patreon.
The original version of Sailfish, an RNA-Seq quantification tool, used minimal perfect hash functions to replace k-mers with unique integers. (The current version appears to be using a Cuckoo hashmap instead.)
This is my attempt to explain how a minimal perfect hash function could be built. The algorithm described here is not exactly the same as the one Sailfish used, but it follows the same idea.
Sections:
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
What is metagenomics and how is it different from phylotyping?
What is Kraken and how can it be faster than BLAST?
Let’s try to sort this out.
Sections:
Links:
Correction: in this episode, I incorrectly state that Kraken operates on phylogenetic trees, whereas in fact it operates on taxonomic trees.
In practice this means that when Kraken cannot decide among several species, it assigns the read to their lowest common taxon (genus, family etc.), not their latest common evolutionary ancestor.
If you enjoyed this episode, please consider supporting the podcast on Patreon.
I talk about allele-specific expression: why it arises and how it can be reliably detected.
Sections:
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Thom Quinn and I explore different ways to transform and analyze relative data arising in genomics.
We also discuss propr, Thom’s R package to compute various proportionality measures.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, I meet with Georg Stricker and Julien Gagneur from the Technical University of Munich to discuss ChIP-seq data analysis and their tool, GenoGAM.
Links:
Register for the Summer School in Bioinformatics & NGS Data Analysis (#NGSchool2017) (Poland, September 2017)
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode Antonio Marco talks about miRNA target site prediction and his tool, seedVicious.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode Aleksandra Kolodziejczyk talks about single-cell RNA sequencing.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
In this episode, Mingfu Shao talks about Scallop, an accurate reference-based transcript assembler.
Links:
If you enjoyed this episode, please consider supporting the podcast on Patreon.
En liten tjänst av I'm With Friends. Finns även på engelska.