FEATHER

FEATHER: a Framework for Experimental Analysis of Tissue and Electrophysiology for Heterogeneous Experiments and Researchers

This project is a MATLAB toolbox developed in the Institute for Auditory Neuroscience to process the various types of data generated in connection to an animal experiment in the institute from in-vivo electrophysiology recordings to results from immunohistochemical analysis of cochlea tissue. The current version only works with the institute specific data types, but future releases should try to include more open source and standardized data formats.

Authors

The FEATHER framework was established and architected by Anna Vavakou. Elisabeth Koert and Niels Albrecht later joined as developers and contributed substantially to the continued development of the toolbox. Elisabeth Koert prepared this public release from the institute’s internal codebase.

Affiliation

Institute for Auditory Neuroscience
University Medical Center Goettingen
Robert-Koch-Str. 40
37075 Goettingen
Germany

---

Toolbox Overview

FEATHER is organized around a few core object types that cover the full workflow surrounding one animal experiment. GUIs allow interaction with the objects to add necessary manual user input for each experiment. FEATHER separates raw and processed data, preserving raw data integrity while persistently storing processed outputs to avoid recomputation. This enables reproducible and efficient workflows across multiple users.

Objects

anex (animal-experiment object)

anex is the central container for one animal experiment.
It links metadata (experiment ID, species, experimenter/user) with raw and processed data directories, and gives an overview of which recordings and analyses are available for that experiment.

At this level, FEATHER enables:

• organizing and loading experiment data consistently,
• listing available ABR, IC, and histology datasets,
• launching standard experiment-wide analyses (e.g., threshold determination across all IC or ABR recordings, summarizing histology results from all image sets).

berabr (Auditory Brainstem Response object)

berabr represents one auditory brainstem response (ABR) measurement recorded with the BERA custom MATLAB software used in the IAN. It stores raw traces, processed FEATHER traces, stimulus information, the raw data directory, and optional calibration information. Multiple berabrobjects can be associated to one anex.

At this level, FEATHER enables:

• ABR preprocessing, and quality handling,
• threshold extraction and waveform-based metrics,
• ABR-specific plotting and calibration-aware interpretation.

associated GUIs for manual user input

exploreBerabr.mlapp together with berabrWaveGUI2.mlapp allow for the manual inspection of all berabr traces associated to an anex and to detect and store the peaks for the different stimuli and recordings.

userberabrOD.mlapp requests user input on the used hardware and optical density filters used for each berabr associated to an anex which is necessary for correctly reading in calibration files.

icme (inferior colliculus multielectrode recording object)

icme represents one inferior-colliculus multielectrode recording. It currently handles data recorded using a 32 channel-NeuroNexus probe and Cheetah recording software. Stimuli are generated with the custom MATLAB software ExpControl used at the IAN. It contains recording metadata, stimulus definitions and calibration values, the raw data directory, and the spike-list with multi-unit activity extracted from raw data for downstream analysis. Multiple icmeobjects can be associated to one anex.

At this level, FEATHER enables:

• extraction/import of multi-unit spike data,
• response analysis (spike rate, PSTH, d-prime, temporal precision, spread of excitation, tonotopy),
• generation of standard IC visualizations for individual recordings (heatmaps, raster-plot, PSTH).

associated GUIs for manual user input

ICuserInput.mlapp requests user input on the used hardware and optical density filters, recording quality as well as stimulation positions within the cochlea used for each icme associated to an anex.

histimg (histology image object)

histimg represents the results obtained from one histology image set (typically one cochlear region / turn).
It stores image metadata and loads quantified outputs (such as cell counts, volumes, densities, and transduction rate) from our custom made, Arivis based histology analysis pipeline that runs on confocal cochlea images (described in Thirumalai et al. 2025 doi:10.7150/thno.104474). Multiple histimgobjects can be associated to one anex.

At this level, FEATHER enables:

• importing histology quantification results from standardized .csv sheets

associated GUIs for manual user input

chooseHistImgToUse.mlapp requests user input to define which histimg should be used when analysing across the full anex in case multiple images have been obtained from the same region.

Surrounding Utility Modules

dirManagement

Helper functions for:

• consistent path generation,
• processed-data folder management,
• cache state handling,
• safety checks to avoid writing into raw/archive domains.

plotFunctions

Reusable plotting helpers used across ABR, IC, and summary analyses to create for quickly visualizing results.

multipleAnexFunctions

Cross-experiment helper functions for pooling results from multiple animals (for example, pooled PSTH-style analyses across multiple anex objects).

Directory Management and Processed Data Layout

As basic infrastructure, FEATHER needs three path settings in the MATLAB session (example initialization shown in the testingScripts):

• Raw data map/drive via ukonmap.
• Processed data map/drive via processedDataMap.
• Processed data base directory via processedDataDirPath.

For a given experiment, FEATHER resolves the processed experiment folder based on the userID of the person running the analysis and the ExperimenterID of the person performing the animal experiment and the ExpID as the identifier of this specific animal experiment (in the IAN <species><ExperimenterID><AnimalID> eg. mav123456):

<processedDataMap>/<processedDataDirPath>/<userID>data/<ExperimenterID>/f_<ExpID>

and raw-data lookups are resolved from:

<ukonmap>/<rawDataDir segments...>

When an anex is initialized, FEATHER creates the processed data folder for this animal experiment and stores the anex object there as:

• E_<ExpID>.mat

For adding analysis-specific comments a KIWI file can be created using initKiwi(anex) that serves as a notepad.

As additional objects are created/processed, files are stored in a consistent structure:

f_<ExpID>/
  E_<ExpID>.mat                     # anex object (experiment-level container)
  <ExpID>_kiwi.m                    # notepad file

  B_<ExpID>_<SeriesID>.mat          # berabr objects (ABR)
  W_<ExpID>_<SeriesID>.mat          # detected waves/ peaks for the berabr object
  List_ABR_raw.mat                  # cached ABR raw list
  List_ABR.mat                      # cached ABR processed list
  ODui_<ExpID>.mat                  # user input table with all berabr associated info
  *.mat                             # additional anex wide ABR analysis results (eg. thresholds)

  List_IC_raw.mat                   # cached IC raw list
  List_IC.mat                       # cached IC processed list
  *.mat                             # additional anex wide IC analysis results  (eg. thresholds)



  HISTO/
    H_<ExpID>_<SeriesID>.mat        # histimg objects
    List_Hist_raw.mat               # cached histology raw list
    List_Hist.mat                   # cached histology processed list
    HistoRes.mat                    # anex wide summary of the histology results across all cochlea turns
    HistoUserInput_<ExpID>.mat      # user input table with all histimg associated info

  ICME/
    IC/
      IC_<ExpID>_<SeriesID>.mat     # icme objects
    RESORT/
      <ExpID>_<SeriesID>_Resort.txt # files containing the metadata, analysis parameters and extracted multiunit activity as the spike-list in a more standardized and easy read in format (if generated)
    SR/
      SR_<ExpID>_<SeriesID>_<t_start>_<t_stop>.mat    # calculated spike rates in specific time windows that can be loaded to avoid recalculating every time (if generated)
  ICUserInput_<ExpID>.mat           # user input table with all icme associated info
  <SeriesID>_tonotopy_res_*.mat     # tonotopy results for acoustic recordings with different analysis methods indicated at the *