• ACA Structural Sciences Society talks
• Awesome lists of interest
• bash
• Bayesian Data Analysis
• Clojure
• Cryocrystallography
• Crystallization
• Crystallographic computing
  • Diffraction image processing
  • SSRL-SMB bash functions
  • Wilson plots
• Emacs
  • Talks about Emacs
  • elfeedorg
  • mooerslab-pdf-auto-tagger-el
• Experimental design
• Grant writing
• Jupyter and Colab
• LaTeX
• Literate programming and computational narratives
• [Machine learning]
• [Molecular Biophysics]
• Molecular Graphics Related
• Molecular Simulation
• [Protein structure]
• PyMOL
• Python
• [RNA Editing]
• RNA Structure
• SAXS
• SSRL talks
• [Structure-based drug design]
• Supercomputing
• Time management
• Typst
• vim/neovim
• Writing tools
  • Manuscript writing support
  • Metacognition tool
  • Accountability tools Includes a cool heatmap.
• Voice computing
  • biological-crystallography-voice-in
• Videos of talks

• Snippets in Jupyter Notebooks, OKC Python Meetup, 10 May 2019.
• Why develop a snippet library for Jupyter in your subject domain? JupyterCon 2020.
• Reproducible molecular graphics with Org-mode, EmacsConf 2021.
• Edit live Jupyter notebook cells with Emacs, EmacsConf 2022.
• Enhancing productivity with voice computing, EmacsConf 2023.
• Managing multiple writing projects, Data Science Workshop, Jan. 2024.
• Voice Computing with Python in Jupyter Notebooks PyTexas 2024, Austin, Texas: April 21, 2024.

This site contains the public repositories of the Dr. Blaine Mooers Lab, Department of Biochemistry and Physiology, College of Medicine, University of Oklahoma Health Sciences, Oklahoma City, Oklahoma, USA. My lab determines the structures of biological macromolecules and their complexes with drugs, mainly using X-ray crystallography and sometimes small-angle X-ray scattering, a method that does not require crystals. We grow crystals using advanced experimental design methods and collect X-ray diffraction data in-house and frequently at the Stanford Synchrotron Radiation LightSource and sometimes at the four other significant light sources in the US. We also do structure-based drug design using supercomputers.

We are not computer scientists; we leave the heavy-duty coding to the professionals. However, we write code occasionally to help advance our science and others'. We have developed software tools to occasionally ease the writing of PyMOL scripts to make molecular images. These tools are available for the top 19 text editors and the Jupyter, Colab, and R Markdown notebooks. These text editors can make your work lighter and have many excellent features, but Emacs is our favorite because, thanks to Emacs Lisp, it is the most customizable text editor. However, we caution that one must take a disciplined approach to Emacs because it is a tinker's paradise.

We often use the web service called Overleaf to write manuscripts, grant applications, slideshows, posters, and lab notebooks in LaTeX. We spent five years trying alternative Markup languages. This was before I learned about Overleaf, which made working with LaTeX much easier. Some of our LaTeX templates are in the repositories found here (see the section on LaTeX below).

Overleaf is a great way to get started with LaTeX. However, Emacs offers more powerful editing features. You can use GhostText to send a page from Overleaf to Emacs for editing. Nonetheless, Overleaf is excellent for collaborative writing. In Chrome or Microsoft Edge, you can use the Voice In plugin to dictate prose in Overleaf.

We are also fans of Jupyter Notebooks. Something inexplicable about Jupyter Notebooks makes coding in them fun. We have developed tools for working on Jupyter with PyMOL and CCTBX. Jupyter Notebooks may be frustrating for those without at least some formal training in computing. You must always be aware of your computer's state, or at least the state you expected it to be in. Those who lack this awareness will run into a lot of trouble.

We also like R Markdown notebooks. Although somewhat less interactive than Jupyter notebooks, these are simpler to put under version control, and they support custom libraries of code snippets with tab triggers and tab stops. The RStudio IDE provides excellent support for R Markdown notebooks. You can have the best of both worlds by editing R Markdown notebooks in Emacs and using the reticulate package to run Python in Jupyter kernels inside the notebook.

We code mainly in Python but sometimes in Bash, R, Clojure, Elisp, Julia, Fortran, and C/C++. Being a polyglot programmer is hard due to limited mental bandwidth.

Our repositories are grouped by category. Some repositories are listed in several categories. This listing is still a work in progress. Some repositories still need to be added to this page. Seventeen repos are forks of other people's repositories. Visit the repositories tab to see the complete list of repositories.

In the spring of 2023, I started a MooersLab site on Codeberg. Codeberg is funded by a foundation and has 13 years of funding. Some GitHub repositories are mirrored on Codeberg.

If you like a repository, please click on the star in the upper right.

• ACA2021
• ACA2022
• ACA2023
• ACA2024

• Bash functions for Schooner at OSCER
• Multiple file transfer
• Function that counts lines of uncommented code

• Bayesian Data Analysis (BDA) Speech-to-Text expansions of acronyms
• JointProb R Markdown and Jupyter notebooks for the Saturday JointProb study group active in 2022 and 2023.

• Lab website with blog posts

• ClojConj 2023 talk about protein structure superposition via probabilistic programming in Clojure
• quizzes about Clojure These quizzes are useful for improving command recall after a break from using Clojure for several months.

Tools to enhance success at cryopreserving protein and nucleic acid crystals.

• Easing cryosolution preparation

Tools to support protein and nucleic acid crystallization.

• Crystal scoring key
• Crystallization drop observation record for 24-well tray
• Vary One-Factor-at-a-Time (OFAT) experimental designs
• Definitive Screening Designs

• aimless summary table extraction
• collection of SSRL-SMB bash functions

• Wilson plot with Python and truncate log file

• Jupyterlab cctbx snippets CCTBX snippets for JupyterLab with the jupyterlab-snippets extension or the jupyterlab-snippets-mutlimenus extension.
• Jupyterlab cctbx plus snippets The variant of the jupyterlabcctbxsnips library with comments to guide editing of the snippets.
• Tagged cctbx snippets The variant for the Elyra-snippet extension for JupyterLab. The snippets have tags in their metadata that you can use to retrieve code snippets.
• Tagged cctbx plus snippets The variant for the Elyra-snippet extension for Jupyterlab. The tab stops in a comment block guide editing the live snippet. Snippets have tags in their metadata that can be used to retrieve them.
• Colab cctbx snippets Colab snippets.

• cctbxsnips-Emacs CCTBX snippets for the yasnippet snippet system in for Emacs.
• cctbxsnips-SublimeText3 CCTBX snippets for Sublime Text 3 (ST3).
• cctbxsnips-VSC CCTBX snippets for Visual Studio Code (VSC).
• cctbxsnips-UltiSnips CCTBX snippets for Vim or NeoVim via UltiSnips plugin.
• cctbxsnips-neosnippets CCTBX snippets for Vim or NeoVim via neosnippets plugin.
• cctbxsnips-Snipmate CCTBX snippets for Vim or NeoVim via the snipmate plugin.
• cctbxsnips-Atom CCTBX snippets for Atom. Have the same format required for Pulsar.

• List of curve-fitting software This list includes both proprietary and free software.

Today's Emacs is not your grandfather's Emacs. The median age of an Emacs user is 32, not 77, as you might expect for a text editing program that is about 40 years old. This is due to a very active community of young users centered around several YouTube channels, Emacs Lisp being well-designed for customizing Emacs, an annual international conference (emacsconf), the popularity of org-mode, and a welcoming community of developers. Elisp was first used in Emacs 19 in 1985, with the GNU Emacs 19.3, about ten years after the project's start.

Emacs is single-threaded. This is a significant limitation when loading large embedded images. The developers have optimized and continue to optimize the part of the codebase that relies on C for speed, so single-threading is generally not a hindrance for most uses.

One group has started a 'lem' project that uses Common Lisp to build a multithreaded text editor similar to Emacs. Unfortunately, lem is not user-ready in our experience. Perhaps Emacs users will migrate to lem vcccccccv. In the meantime, it is worth continuing to use Emacs, knowing that a multithreading alternative is under development.

The Emacs project is open-source and run by several hundred volunteer core developers. The project runs by consensus. There is no succession issue. The project's benevolent dictator for life stepped down about 15 years ago. After doing so, an explosion of creativity occurred.

According to the Lindy effect, it is likely that Emacs will remain in use for another 50 years. Exposure to Elisp coding in Emacs leads some to pick up other languages in the Lisp family, including Clojure. This, in turn, leads them to understand programming more deeply.

Mastery is in the eye of the beholder. Total mastery is impossible with >5000 packages.

• Popup menus of GNU Emacs core commands you want to master
• Configuration for Emacs30 to support structured editing
• Some simple hydras in Codeberg repo
• Quizzes about Emacs to improve recall of keybindings Ten minutes boosting your recall on a particular topic can save 10, 100, or 1000 hours of Google searches, especially if you are prone to getting sidetracked in the Google-verse.

These files can help you on your Emacs journey. You probably just want to copy bits of configuration to integrate into your own workflow.

• A simpler approach to using org in place of the init.el file
• Compiling Emacs29 on Sequoia (macOS 15.01) with tree-sitter
• Compiling Emacs30 on Ventura (macOS 13.2) with tree-sitter. This protocol compiles Emacs from source with tree-sitter support for structured code editing on macOS.
• Compiling Emacs30 on Ubuntu22.04 LTS with tree-sitter This protocol compiles Emacs with tree-sitter support on Ubuntu.
• crafted-emacs user config My user configuration file (config.el) for the crafted-emacs profile. I installed use-package and used it to install several packages. The base profile has the Doom theme, Vertico completion stack, and eglot. I added my org-agenda config and my favorite key bindings. I am currently luvin it.
• latex-emacs29 configuration A configuration file enhancing the use of LaTeX in GNU Emacs version 29.0.5. Includes org-mode configuration.
• latex-emacs28 configuration Similar to above but for GNU Emacs version 28.
• latex-emacs30 configuration Similar to above but for GNU Emacs version 30.
• scimax user.el Scimax is the Emacs configuration that Dr. John Kitchin of Carniege-Mellon University is developing to support reproducible research by scientists with GNU Emacs. The user.el file for Scimax lets users add keybindings and packages. I added features from latex-emacs and config.org.
• config.org My default GNU Emacs version 30 configuration file.

Org-mode blends the best features of Markdown and LaTeX. Although you can use Org mode without knowing LaTeX, you don't have to give up any of your LaTeX skills and knowledge to use Org mode. You gain powerful tools for navigating the file, making major rearrangements of its subtrees, and interfacing with many other programs, such as Gnuplot, to generate data plots from tables. Org-mode's table editor is reason enough to adopt its use because it makes assembling a table easy. For example, you can draft a table in Org mode, drop it into Copilot, and ask it to reformat it in LaTeX.

• Book project in org-mode
• Writing log template in Org-mode Template for tracking your actions, decisions, and plans for any serious writing project that takes more than one session to finish. The LaTeX version is listed in the LaTeX section for use on Overleaf or in Emacs.
• Org-mode manuscript template The manuscript will be exported to PDF after being run through LaTeX. This is for the first submission. This is for the lovers of org-mode. The export from org to PDF is fast and painless in Emacs.
• Basic modular annotated bibliography in org-mode
• Full-featured annotated bibliography in org-mode

See the section about LaTeX for more repositories.

• Slides about workflow in LaTeX These were presented to the Berlin Emacs Meetup in August 2022. Regrettably, the talk was not recorded. It was a 90-minute presentation.
• snippets for latex-mode in Emacs LaTeX code snippets for yasnippets.

The more I am exposed to other Emacs community members, the more I appreciate how little I know about Emacs. The growing disease of expertise tends to inhibit sharing knowledge with beginners. Advanced beginners tend to give talks that are more accessible to beginners because they tend to assume less prior knowledge. As an advanced beginner, I find the pressure of assembling a talk an effective way of forcing myself to master new material.

• emacsconf 2021 talk about rendering molecular graphics with PyMOL in org literate programming files Video and comments.
• EmacsATX Meetup Talk, 4 May 2022: functional programming in Emacs lisp Slides.
• Video link to talk about GhostText, Oklahoma Data Science Workshop, July 2022
• Slides from talk about GhostText, Data Science Workshop, July 2022 GhostText rocks! It enables Emacs to edit text boxes on websites. You can use GhostText to bring the full power of Emacs to Overleaf (Important: use the legacy version rendering of the text area in Overleaf to get text imported into a LaTeX buffer).
• emacsconf 2022 talk about using Emacs to edit Jupyter cells via GhostText and atomic-chrome Video and comments.
• emacsconf 2023 talk about using voice-computing with Emacs to enhance productivity Video, transcript, and comments.
• DSW November 2024: A Science Writing Workflow in Org Mode: slides and link to talk
• emacsconf 2024 talk about managing science writing metadata
• emacsconf 2026 talk about developing an antifragile writing habit

• Voice-In library for Emacs
• Voice-In library for org-mode

You can run PyMOL in Jupyter via PyMOL's Python API.

Multiple Jupyter extensions support snippets. However, the support does not include tab stops or tab triggers, two standard features of snippet support in modern text editors. Libraries with plus at the end of their name include a second copy of the code in a comment. This copy is marked with the tab-stop locations where you might want to change the parameter value from its default.

PyMOL Python snippets for use in the Classic Jupyter Notebook with the snippets_menu notebook extension.

• jupyternbclassicpymolpysnips
• jupyternbclassicpymolpysnipsplus

PyMOL Python snippets for JupyterLab using the jupyterlab-snippets extension.

• jupyterlabpymolpysnips PyMOL Python snippets for use in JupyterLab with the jupyterlab-snippets extension.
• jupyternbclassicpymolpysnipsplus

PyMOL Python snippets for use with the elyra-snippets extension and with PyMOL. You can retrieve these snippets using tags in the elyra-snippet GUI.

• taggedpymolpysnips
• taggedpymolpysnipspymolpysnipsplus

These editors work in Jupyter Notebooks via the GhostText browser plugin.

• jupyter-vsc-pymolpysnips
• jupyter-st3-pymolpysnips
• jupyter-emacs-pymolpysnips
• jupyter-ultisnips-pymolpysnips
• jupyter-snipmate-pymolpysnips
• jupyter-neosnippets-pymolpysnips
• jupyter-atom-pymolpysnips

• emacsconf 2022 talk about using Emacs to edit Jupyter cells via GhostText and atomic-chrome

• colabOpenSourcePyMOLpySnips PyMOL Python snippets for use in Colab with open source PyMOL.
• colabPyMOLpySnips PyMOL Python snippets for use in Colab with the incentive PyMOL.

• Jupyterlab cctbx snippets CCTBX snippets for JupyterLab with the jupyterlab-snippets extension or the jupyterlab-snippets-mutlimenus extension.
• Jupyterlab cctbx plus snippets The variant of the jupyterlabcctbxsnips library with comments to guide editing of the snippets.
• Colab cctbx snippets Colab snippets.

We provide easy and fast-to-deploy experimental designs for crystallization setups in Excel workbooks. These designs take three minutes to customize for your experimental needs. Plots of the results against each factor level are automatically generated, saving users many hours of tedious coding for each crystallization tray. These workbooks could be adapted to other kinds of experiments.

The OFAT experiments are often used to screen for active factors. The active factors can then be used in DSD designs to find the optimal conditions to return large crystals. OFAT experiments are inappropriate for this objective because they do not measure interactions between factors. Factor interactions are common in protein crystallization.

DSDs are typically limited to three active factors because of their small sample size and weaker statistical power. Use larger traditional response surface method designs (RSM) if more factors are suspected to be active or if two or more interactions are expected. DSD can be thought of as a subset of RSMs. Their small size is attractive for protein crystallization, where the sample is expensive and scarce. The small size of DSD experiments makes it possible to replicate them to estimate variance.

• Vary One-Factor-at-a-Time (OFAT) experimental designs
• Definitive Screening Designs

• Design of Experiments (DoE): doe.csv

• Voice Commands for NIH grants
• Voice commands for NSF grants

Until recently, I wrote a lot in Emacs on 750words.com via GhostText. I configured the atomic-chrome package in Emacs to switch to LaTeX mode when GhostText connects to 750words. I can also use Emacs to write and edit text in Overleaf via atomic-chrome and GhostText. This is a very cool way to extend Emacs to websites with weaker text-editing support.

Now I use voice-to-text to dictate in 750words.com or Overleaf. I use voice commands to insert snippets containing LaTeX code. See the section below.

If you don't have extensive experience writing in a markup language like HTML or Markdown, start with a Markdown flavor to become productive as quickly as possible. Many ideas behind the typesetting program Markdown will transfer to LaTeX. Markdown is easy to use and highly intuitive. Mastering Markdown provides a fallback system if you get stuck debugging errors in your LaTeX document. This is a common experience for beginning LaTeX users.

MyST expands to Markedly Structured Text. MyST is an enhanced version of Markdown developed to interface with Jupyter notebooks and support their utilization in scientific publishing. You can export files to LaTeX, MS Word, and PDF.

MyST is part of the dream of converting Jupyter into a literate programming platform and supporting the conversion of scientific Publications into highly extensible documents. You can edit MyST files in the newer versions of JupyterLab and any text editor.

It provides a preview in your default browser that is updated each time you save the main.md file. HTML lovers are wowed by this feature.

Live preview extensions are available for Emacs, Vim, and VS Code. It feels like an underdeveloped version of org-mode, which had all the export capabilities over a decade ago. It may be more analogous to the well-planned, lean, and efficient R Markdown for the R programming world.

These markup languages are more potent than Markdown. However, very few publishers will accept the source files for these markup languages. You may be able to submit the PDF generated by one of these systems on your first submission. However, you will still need to transfer the contents of your file to MS Word or LaTeX.

A nice introduction to Typst in October 2024 by Dr. Chase Brown of the University of Central Florida is found here.

This typesetting language uses BibTeX and LaTeX Math. RestructuredText is used by some in the Python world to generate documentation. The annual SciPy Proceedings have used it for many years. It harnesses indentation, for better or worse. It is not hard to use for a LaTeX user.

Typst compiles many times faster than LaTeX. Typst provides a nearly live preview and thus instant gratification. This may be the main reason to use it for simple documents.

Typst has a more modern bibliography management system, but it also supports using BibLaTeX, which will ease adoption by new users.

Typst also utilizes a more computer programmer-oriented approach to typesetting by using functions. Typst also provides boilerplate styling templates similar to classes in LaTeX. Math typesetting is similar to LaTeX, so LaTeX users will find it easy to adopt. Typst set out to replace LaTeX but retains many features.

It is easy to write a generic manuscript in typst for the first submission PDF. The upside is how quickly the document compiles. The corresponding writing project log file template in typst is found here. The corresponding modular annotated bibliography template in typst is found here.

Typst has been integrated into org-mode. You can use a #+typst: keyword to start lines with typst commands. There is even a typst submenu for exporting to PDF or typ files.

An abandoned typst-mode project exists on GitHub. The author shifted his efforts to the typst-ts-mode project. Unfortunately, typst-ts-mode package is harder to install.

Oh well, typst-mode is much easier to install and use. It provides syntax highlighting, and entering C-c C-p compiles the document to PDF and instantly displays it in your default PDF viewer. That is all of the support that I need most of the time.

It might still be easier for computer programmers to adopt typst than for nonprogrammers. Writers who are not computer programmers will quickly master Common Markdown. The next step in markup will probably still be the LyX editor, which offers a what you see is what you get approach to using LaTeX.

• LaTeX bare-bones template file This template is for people who want to make a simple document without a lot of bells and whistles.
• LaTeX tips
• Configuration for Emacs30 to support structured editing
• latex-emacs29 configuration A configuration file enhancing the use of LaTeX in GNU Emacs version 29. This is the current release version of GNU Emacs. Includes org-mode configuration.
• latex-emacs28 configuration Similar to above but for GNU Emacs version 28.
• latex-emacs30 configuration Similar to above but for GNU Emacs version 30.
• Slides about workflow in LaTeX presented to the Berlin Emacs Meetup August 2022. Not recorded. It was a 90-minute lecture.
• slideshowTemplateInLaTeX Slideshow template to use with the LaTeX package Beamer that supports the making of slides. Our templates make slides that don't look like they were made in LaTeX because we hide the navigation icons no one uses and replace the triangles in bulleted lists with round bullets. Our slides look like they were made in PowerPoint.
• posterInLaTeX uses Beamer to make a poster with a simple design. This lets you whip up a poster in a few hours. It is much easier than using a Gaint slide in PowerPoint. Our approach is also simpler than the default way to make a poster with Beamer.
• LaTeX manuscript template Generic template for the first submission as a PDF.
• Writing log template in LaTeX Place to track progress and plans behind a manuscript.
• Annotated bibliography template in LaTeX Every writing project needs one of these. These should be utilized more. They are good for recalling a paper's relevant features.
• ditto but for org-mode
• Diary for 2024 in LaTeX Designed for use on Overleaf.com but can compile locally. Each day is a section. Each month is a chapter. Has an automatically generated index (most Markdown cannot generate indices). Very feature-rich and ready to use. Makes a great electronic scientific notebook because it is searchable and indexed. > A 1000-page document at the end of the year compiles to PDF in under a minute on Overleaf.
• snippets for latex-mode in Emacs My LaTeX code snippets for yasnippets.
• The writer's creed Guidelines for greater productivity as a writer.
• bib2item3 Python script to convert bibtex.bib file to bibitems in tex file format. Bibitems are required by some publishers. They are a pain to create manually. - LaTeX Voice In Plus library Speech-to-text commands for inserting LaTeX markup.

See the sections on Org-mode, Jupyter, and Colab.

Collection of links for a 15-hour course on Molecular Graphics

Coming soon. Includes Structure-based Drug Design.

• EasyPyMOL: go here for an easy start for interactive horizontal scripting on the command line in PyMOL and for the roundview.py function. YouTube videos show how to install the software on various operating systems.
• pymolshortcuts: These are very popular. You don't need to be a coder to use them. This is what most people will want to install. The README.md explains the installation and use.
• pymolsnips: For people writing traditional vertical *.pml PyMOL scripts in a text editor. The README.md describes how to install each of the 17 supported text editors. It then shows how to install the snippet library in each editor. Animations show the snippet library in use in each text editor.
• pymolpysnips: For people wanting to run PyMOL in one of these computational notebooks: Jupyter, Colab, RStudio, and Org-mode.
• PyMOL wall hangings: Protocol to convert molecular artwork into large images for framing and hanging on a wall.

• pymolrc My pymolrc file.
• pymolsnips A large collection of PyMOL macro language (pml) code snippets. It links to a webpage with extensive instructions on installing 19 text editors.
• pymolshortcuts A large collection of functions mapped to aliases or . This collection of functions is very useful for non-coders.
• EasyPyMOL Support for horizontal scripting in PyMOL.
• orgpymolpysnips Support for generating computational narratives (literate programming plus interactive computing) in org-mode with PyMOL.
• rstudiopymolpysnips Support for literate programming in RStudio with PyMOL.
• taggedpymolpysnips Support for literate programming in JupyterLab with tagged snippets with the elyra-snippets extension and with PyMOL.
• jupyterlabpymolpysnips PyMOL Python snippets for use in JupyterLab with the jupyterlab-snippets extension.
• jupyternbclassicpymolpysnips PyMOL Python snippets for use in the Classic Jupyter Notebook (invoked with jupyter nbclassic with Jupyter Notebook version 7).
• colabOpenSourcePyMOLpySnips PyMOL Python snippets for use in Colab with open source PyMOL.
• colabPyMOLpySnips PyMOL Python snippets for use in Colab with the incentive version of PyMOL.
• PyMOL wall hangings Protocol for making oversized images framed and hung on a wall.

See the sections on PyMOL, CCTBX, and Jupyter and Colab. Python is in many repositories.

• PyTexas talk: Voice computing with Python in Jupyter notebooks (slides and audio), 2024 Apr. 21

• rstudiopymolpysnips Support for literate programming in RStudio with Python code for PyMOL.

• Count atoms in RNA by element

• SmallAngleScatteringWebpage

Talks at the SSRL/LCLS Users Meeting:

• 2024 Workshop talk: Tools to Ease the Use of PyMOL

• Gantt chart Support mutliple project management.
• TimeBlocks Form for daily scheduling with time blocks.
• TimeSpent Time tracking by project with Google Sheets.

High-performance computing (HPC), cluster computing, and cloud computing.

• Bash functions for Schooner at OSCER
• Emacs configuration file for schooner supercomputer
• Multiple file transfer with scp
• Vim configuration file for the schooner supercomputer at OU-Norman
• Vim configuration file for SSRL SMB cluster
• Tutorial for biologists on using the supercomputer at OU-Norman

• Vim configuration file Current .vimrc file for Mac OSX 10.15
• Vim configuration file for schooner supercomputer
• Vim configuration file for SSRL SMB cluster
• Neovim configuration file
• qvim Quiz about Vim commands to improve your recall of the commands after a break from using Vim.

Voice computing can be divided into three activities:

• speech-to-text (dictation)
• speech-to-commands
• speech-to-code

Speech-to-text is the easiest to master, the most widely applicable, and the most effective way of enhancing academic productivity.

The video below shows how to use a voice trigger to insert a chunk of LaTeX code I use to plan each day. It inserts after you say insert start day. You can easily make a Markdown variant customized to meet your needs. I use Voice In Plus in Google Chrome. It also works in Microsoft Edge. It uses the browser's built-in Speech-to-text software on your local machine, so latency isn't much of an issue.

• Software that I use
• Talon Voice for voice control and computing
• Voice In Plus for dictation and voice control in the web
• Whisper for dictation, includes my text replacements
• Voice Typing in Google Docs for dictation
• Words counts before and after adopting
• Voice computing-related talks

My sampling of the available -related software is highly skewed towards the Mac operating system and Google Chrome. I have yet to complete a comprehensive survey of the software available for Windows, Linux, and other browsers. If you use these systems, the review below will give you a taste of what may be available for these other operating systems.

Voice Typing is widely available through Google Docs. Voice Typing supports almost 100 voice commands but does not support custom commands.

Voice In Plus has accuracy similar to Voice Typing, with the addition of custom commands. I use the latter every day to dictate the first drafts. I rewrite using the keyboard.

OpenAI's Whisper may be an order of magnitude more accurate. It has a latency issue, so live transcription and interactive editing are difficult. It is also not easy to fine-tune with custom commands.

My primary use of Whisper is for transcribing audio files. I run Python scripts to correct the initial transcript. These corrections include expanding contractions. This code also supports inserting simple commands like new paragraph during dictation.

A very new use of Whisper for me is inside Emacs. After my third attempt, I finally figured out how to run Whisper in Emacs via the whisper.el package. The package automates compiling the whisper.cpp software and downloading the Whisper LLM. You record a local audio file of your dictation, then transcribe it into text that appears in the current Emacs buffer. The M-H r keybinding starts the audio recording, and the C-u M-x r keybinding transcribes the audio file. Here, H is for the Hyper key; I have it mapped to the right-hand command key (Alt key) on a non-Mac keyboard.

The natural unit of dictation with whisper.el seems to be the paragraph. This form of dictation is less interactive than Voice In Plus, where you can use the keyboard to edit the sentence being dictated. I haven't figured out the Elisp code to apply my text replacements during transcription.

The Superwhisper. app for the Mac is interesting. It supports dictation in e-mail and elsewhere. Its support for custom commands is limited; it still needs to support bulk uploading of text replacements before I am sold. You have to pay a subscription fee to access all its features. Its features may appeal to people with numerous electronic devices that must be synchronized with their primary computer.

Macs with a newer Silicon chip have a local LLM that converts your voice to text. There are two programs, Dictation and Voice Control. The latter program is more interesting because you can add custom voice snippets. On my old Intel-based MacBook Pro, Voice Control did not do accurate dictation. It was also slower because it utilized a remote LLM.

However, with a new M4 chip-based MacBook Pro, Voice Control does a very accurate dictation job. Compared to Voice In Plus, it has longer lag. Voice Control works anywhere you can place a cursor.

Talon Voice supports all three activities, but in my experience, speech-to-text accuracy still needs to catch up to Voice Typing and Voice In Plus. I have been using Talon Voice for the last two activities. I plan to use Talon Voice more for dictation. The Talon Voice software is designed for non-coders. It can be configured with Talonscript files that use a simple syntax, a subset of Python. You can extensively customize Talon Voice using the Talonscript files without writing any Python code.

Voice computing supports extending your productivity when you get tired of typing. Voice computing also lets you issue commands and dictate text while standing. Voice computing could cure the envy of colleagues with standing desks. You can speak commands to your computer while standing 20 feet away from it and looking out the window. You can give your body and eyes a break from sitting and staring at the computer screen while remaining productive.

I saw a three-fold increase in the total number of words captured in September and October 2023 compared to June and July 2023 after I started using speech-to-text in mid-August 2023. I have had the same amount of out-of-town travel during the first two periods. I did not make a conscious effort to capture more words per day.

On the day after Thanksgiving in 2023, I generated almost 12,000 words. From that point to just before Christmas, I made a concerted effort to generate at least 5,000 words daily. Then family obligations reduced my daily word generation. Nonetheless, I had a five-fold increase in the total word count for November and December 2023.

A three-fold increase is more realistic given the disruptions from experiments, administration, teaching, service, and work-related travel. The slight decrease in word count after the new year 2024 may reflect a shift in my focus toward turning more of my writing into publishable products. This decrease may reflect more effort focused on editing text.

In November-December 2024, I spent 80 hours preparing a talk for the Data Science Workshop and 140 hours preparing an Emacsconf talk. This helps explain the decline in word count in November. I started posting the daily writing effort in December. This might have changed my behavior as I tried to finish a manuscript. In any event, the total word count for 2024 was 1,090,627 words. That's equivalent to 12 long novels. Too bad my words are not so golden!

I may have compensated by using a digital voice recorder (DVR) in mid-January 2024 while away from my computer. The DVR is better than a cell phone for recording speech because you don't have to hold it next to your mouth. The DVR can record a Zoom meeting while lying on a table or a seminar from the back of a room. Use it ethically. I transcribe the audio recordings with a locally installed version of OpenAI's Whisper. I use Python and bash scripts to reformat the text so each sentence is on its own line for easier downstream editing.

I've been using this device daily since late January 2026. It is the best $85 investment I have made in a long time. It is worth ten times as much.

Some of my mutterings are rubbish that I cut from the transcripts. I occasionally harvest valuable thoughts that are worth retaining. I also use the DVR to paraphrase technical documents that would otherwise be tedious to read. The DVR has extended my generative writing time by 15-30 minutes daily.

Spending more time on rewriting will generate fewer new words if the time spent writing stays constant. Rewriting is very time-consuming; it can take as long as, or longer than, generating the text initially. AI-powered writing assistants make suggestions, but they are not a silver bullet. You still have to spend a lot of time making the right selections; otherwise, your writing will often suffer. These assistants have not saved me time, and they have often wasted it. But if they keep you engaged in rewriting, they can be helpful.

Text written for grant applications and articles will take even more time because it needs to be highly polished. Sufficient time has to be scheduled for it. I plan to spend more time rewriting, so you can expect the number of new words generated in the following bimonthly reports to decline.

The slump from November to February 2025 was due to switching laptop computers. The new computer did not run Whisper correctly at first, which dramatically reduced my ability to convert audio files to text.

The surge in March-April 2025 likely resulted from several factors. First, I discovered the write file command in the whisper.el` package of Emacs. This package uses the C++ version of Whisper. This version is about 100 times faster than the Python version and much easier to install. It is not susceptible to the dependency issues that plague Python.

I run this command in a directory where I have moved the audio file. This directory is the default directory whenever I log into Emacs. I am prompted for the audio file, which is easy to select because it starts with today's date. Audio files are then transcribed in a few seconds into one long line. I select this line in Emacs to make it an active region. I apply a custom-made command (found in mooerslabfunctions.el) that splits the line by sentences, so I have one sentence per line. I either parse the text manually or have a chatbot parse it for me.

Because the final destination of the text will be a LaTeX document, I use the following prompt before I supply the transcript. The to-do list in org lets me paste these to-dos into an org-mode file.

Parse this transcript into paragraphs and subsubsections by topic. Include index keys below the subsubsection headline. Extract the TODO items into an Org-mode-style checklist starting (e.g., - [ ]). Enclose this list in a LaTeX minted block between \begin{minted}{org} and \end{minted}. Return the text in LaTeX. Here is the transcript:

Here is my current variant that exports the to-dos in LaTeX:

Parse this transcript into paragraphs and subsubsections by topic in LaTeX.
Include index keys below the subsubsection headline.
Extract the TODO items into a LaTeX-style itemized list with the following first line: \begin{itemize}[label=\unchecked].
Enclose this list in a LaTeX \end{itemize}
Return the parsed text in LaTeX without the preamble and the \begin{document} and \end{document}.
Correct the grammar and return one sentence per line.
Return the index keys in the format \index{<keyword or phrase>}.
Replace all English contractions with their expansions.
Return the index keys one per line.
Here is the transcript:

I have a chatbot fill in the index keys and reorganize the prose by its identified categories. These categories become the subsection titles. Adding index keys is another distracting chore. These are low-level tasks I don't really like and tend to neglect. Having the chatbot do this labor has been liberating.

Someday, I will take the time to direct a chatbot to go through my old diary entries and clean up the text in a similar fashion. For example, I have a folder of 365 TeX documents for the prior year. I could direct it to review each document and clean up the text in response to the prompt above. I must go back through the rewritten text to correct errors and misinterpretations.

I also have the chatbot extract to-do items and put these into a list in org-mode. I dictate the term `to-do' to mark the beginning of a task for my to-do list. I then copy and paste the list into my daily entry in an org-mode document where I keep my daily plans.

I broke my lower right leg in two places on September 30, 2025. I sheared the tibia and fibula below the knee and smashed into many parts of the shin bone further down my leg. The second site has caused a lot of trouble. Two metal plates and fourteen screws were used to hold the bones together. They were inserted through a 20 cm incision. The initial orthopedic surgery wound became infected during the first two weeks. They removed the infected tissue in five subsequent operations over the next six weeks. Three and a half months after the accident, the metal was removed on January 14 because the incision wound was not closing. One thick metal plate was making the skin on my shin too taut.

The subsequent surgeries delayed my recovery. I was not able to put any weight on my right leg for 10 weeks. By being forced to be very sedentary, my physical stamina diminished greatly. This, in turn, impacted my mental stamina.

I was able to keep up my daily entries at 750 Words, but I couldn't stick to my writing schedule during most of my recovery. I would start my writing schedule multiple times, only to have it interrupted by medical matters. The daily entries included summaries of what I accomplished the day before and reflections on my medical condition. I also kept a record of my drug use. I have had to take three sets of 3 to 4 antibiotics, each set for six weeks. The second set was injected via a PICC line.

Part of the problem was that the painkillers I was taking interfered with my sleep. During the first 10 weeks, there were many days when I was only able to get 2 to 4 hours of sleep. Once I could put some weight on my right leg and move around more, the fatigue from doing so forced me to sleep several hours more than I normally do. Receiving medical treatment actually involved a lot more time than I had expected. I had weekly doctor's office visits because of the seriousness of my infections, and I had multiple weekly visits from in-home care people. One week, I tallied up 50 hours spent on matters related to my injury. The recovery process was essentially a full-time job.

You would think that being forced to be sedentary would be a wonderful opportunity to get a lot of writing done. I had that expectation, but it turned out to be very unrealistic in the presence of serious physical trauma. Although I was on extended sick leave, I kept up with many of my professional obligations by reviewing manuscripts and serving on committees via Zoom.

One alternative is to use the built-in Dictation software for Mac. The Mac dictation software used to be worse than the alternatives, like Voice In. I tried the Mac dictation software again on March 22, 2024, The error rate was reduced enough to make it competitive.

Its advantage is that it can operate anywhere. It works inside Emacs and on webpages, including webmail applications like Outlook and Gmail. Dictation relies on an internet connection on older Intel-based Macs. The generated sound waves are sent to an Apple-run server and then returned as text. I found the delay caused by this transfer to be unnoticeable.

The Voice Control software provided by macOS has not improved since I last tried it. Its error rate is too high to use for dictation. The advantage of Voice Control is that it supports custom commands that can run GUI-based software.

We use OpenAI's Whisper software via the Python openai-whisper package to transcribe my audio recordings with the DVR for almost a year. I took a three-month break after I switched to an Apple Silicon computer in December 2024, and this Python package was tough to install because of complex dependencies. I finally found that I could compile and run whisper.cpp, which is 100 times faster than the Python-based software.

An alternate approach is to play the DVR with Voice In Plus in Google Chrome and use the cursor in a text area in the browser to run the transcription and write the text to the text area. This ties up the computer for the duration of the recording.

I wrote Python code to reformat the Whisper transcript to make it easier to edit. I use Python to make my text replacements in the transcript via regex. My approach with this software is more cumbersome than the straightforward approach in Voice In Plus.

In addition, I reformat the text so that each sentence is written on its own line. This format eases the deletion of lines containing rubbish. You can advance to the following line with the down arrow key in most text editors. If the current line is to be deleted, you enter Control-K.

I have used the text transcribed by Whisper less than text generated with Voice-In Plus. Whisper often inserts garbage during long pauses. It takes a lot of effort to parse the transcribed text.

It is easier to redo the dictation via Voice In and make corrections as you go. Dictating the first draft during the morning commute is analogous to generating a verbal rough draft. If I leave the DVR at home, I will still compose some prose by speaking aloud. Once I've composed the initial draft verbally, it is easier to revise it with better word choices after I reach my office.

Of course, generative writing is a small part of writing. Rewriting, data analysis, figure-making, reading the literature, and managing co-author input take up most of the time. You can try rewriting a small part of a manuscript before a commute by re-reading the part in question immediately before you leave. This sounds boring, but exciting new ideas often emerge during this exercise.

Below are links to programming tools to aid the adoption of voice computing.

• Expand dictated contractions automatically
• Run bash commands in the terminal by voice command
• Open favorite web pages by voice commands
• Master Talon Voice phonetic alphabet with interactive quiz written in Python
• Master Talon Voice with interactive quiz written in Elisp for Emacs

This browser plugin works in Google Chrome and Microsoft Edge. It depends on the browser's Speech-to-Text API. Interactive sessions with it persist for up to 5 to 10 paragraphs, depending on your activity and speed of dictation. If you pause too long, the software will deactivate after 3 minutes. If you dictate too quickly, you will get ahead of the software, which will hang. You might lose text for some of the words you dictated. Nonetheless, this software is superior to Mac dictation software, which will only persist for an average of three sentences because it sends the sound waves to a language model located on a remote server.

\url{https://github.com/MooersLab/MooersLab/assets/15176203/0b543abf-aa3e-4dc9-acb4-e98283e6b4f3}

• Master basic Voice In commands with interactive Python quiz
• Master basic Voice In commands with an interactive quiz in Emacs
• Library of 186 English contractions (a must for nonfiction writing)
• Jupyter notebook voice triggers
• Full library of commands for Voice In Plus
• Org-mode
• OUHS-related acronym expansions
• LaTeX
• Generate new writing project commands
• Books about scientific writing mapped to voice triggers

• Biological crystallography (MX)
• Cryogenic electron microscopy (CryoEM and CryoET)
• Small Angle Scattering (SAS)
• Computational Crystallography Toolbox (cctbx)

• Design of Experiments (DoE)

• Bayesian Data Analysis (BDA)
• Artificial intelligence (AI)
• Software Development
• Markdown in Jupyter notebooks

• Voice-In library for Emacs
• Voice-In library for org-mode

• NIH grants
• NSF grants

This is a standalone application that supports speech-to-commands and speech-to-code. It depends on the presence of an internet connection because it uses remote servers to convert sound waves into text and commands. Serenade runs in web browsers, in the terminal, and at command prompts. It is more versatile than Voice In Plus but less versatile than Talon Voice. It may be a good bridge between Voice In Plus and Talon Voice.

• Expansions of English contractions

• Master Voice Typing commands with interactive quiz

• Bash function to transcribe audio files with the Python version of Whisper
• Emacs package to read audio files directly off DVR during transcription

I have given three talks on this topic: two at the regional level and one at the international level.

• emacsconf23 talk speech-to-text and speech-to-commands, 2023 Dec. 2 on YouTube
• PyTexas talk: Voice computing with Python in Jupyter notebooks (slides and audio), 2024 Apr. 21
• ACA slideshow (no video), 2024 July 8

These are some programming tools designed to support greater writing productivity. There are three elements: progress tracking, support for manuscript writing in LaTeX, and the writing log. The last is very helpful for organizing and managing a writing project. It is analogous to a project-specific laboratory notebook.

• Books about scientific writing mapped to voice triggers

• Gantt chart Support mutliple project management.

• LaTeX classic manuscript template Generic and classic template for the first submission as a PDF. It has enhanced headers to ease sorting pages when stored with two or more manuscripts while traveling.
• Typst classic manuscript template Generic and classic template for the first submission as a PDF. It has enhanced headers to ease sorting pages when stored with two or more manuscripts while traveling.
• Org-mode classic manuscript template Generic and classic template for the first submission as a PDF. It has enhanced headers to ease sorting pages when stored with two or more manuscripts while traveling.
• Annotated bibliography Template in LaTeX Classic and boring variant.
• Modular annotated bibliography with BibTeX in LaTeX Modular annotation files for reuse and to evade mangling by BibTeX. Can be fully illustrated and include multiple paragraphs. Includes index, glossaries, and Literature Cited for references outside the annotated bibliography. Drag and drop zip file available for instant gratification on Overleaf.
• Modular annotated bibliography with BibLaTeX in LaTeX Modular annotation files for reuse and to evade mangling by biber. Can be fully illustrated and include multiple paragraphs. Includes index, glossaries, and Literature Cited for references outside the annotated bibliography. Drag and drop the zip file, which is available for instant gratification on Overleaf.
• bib2item3 Python script to convert bibtex.bib file to bibitems in the TeX file format. Bibitems are required by some publishers. They are tedious to create manually.
• Emacsconf 2022 talk about GhostText on YouTube, December 2022 Includes a demonstration of using Emacs to edit a document stored in Overleaf.
• LaTeX Voice In Plus library of speech-to-text commands

• Writing Log template in LaTeX. This is the favored format. LaTeX beginners can easily use this document on Overleaf without any configuration.
• Writing log template in Org-mode Has all of the features of the LaTeX variant. Favored by many Emacs users. Org-mode is a powerful analog of Markdown that can interpret LaTeX code blocks. If you are a LaTeX user, you can view org-mode as LaTeX with extensions that ease the assembly of tables, support time management via org-agenda, support knowledge management via org-roam, and support interactive computing.
• Writing log template in typst Typst is a new alternative to LaTeX. It is built on Rust and instantly compiles small documents like this writing log.
• Writing log template in reStructuredText reStructuredText is used by programmers for documentation.
• Writing log template in Markdown Markdown variant. Rendered to PDF by most good text editors.
• Writing log template in ODT ODT can be read by OpenOffice, LibreOffice, and MS Word.
• Writing log template in DOCX for MS Word MS Word variant—the least suitable format for this task, in my opinion.

Progress tracking can support momentum and enthusiasm for writing tasks that cannot be completed in one sitting.

• Autogenerated GitHub-style calendar-based heatmap of time spent writing
• Track writing progress in 2024 and 2025 Enables tracking effort on separate projects and generates plots of cumulative effort across projects in bar plots. Updating the records at the end of the day takes only several minutes. Required tracking start and stop times,
• TimeSpent Hourly time tracking by project with Google Sheets. This is useful for making monthly time reports and learning how much various writing projects take. It requires updating every several hours throughout the day. Updating the record can be painful if you fall behind by a day or more. Only some people have the discipline to do time-tracking. If you cannot do it, you are "regressing towards the mean," as Sir Francis Galton would say.
• Writing contract template This contract is in Microsoft Word, Markdown, LaTeX, and Org-mode. This accountability system works only if your partner is similarly committed to daily writing. I had a good run of four years with one person from out of state. I would consider doing this again.

• NIH: R01 CA242845, AI088011
• NIH: P30 CA225520 (PI: R. Mannel)
• NIH P20GM103640 and P30GM145423 (PI: A. West)

MooersLab/MooersLab is a ✨ special ✨ repository because its README.md (this file) appears on his GitHub profile 👋.