Identify code - jungheejung/life-encoding GitHub Wiki

🌱 Purpose of this page: categorize cara's code into 4 categories

  • hyperalignment
  • annotation
  • ridge regression
  • variance partition

👩‍🚀 acronyms

  • aa: between-subject anatomical alignment
  • ha: between-subject hyperalignment
  • ws: within-subject

👏 Identifying and Sorting code

If you've walked through a code, please use the following keywords -- hyperalign annotate ridge variance -- in front of the code!

Deepanshi's To Do

  • Dockerfile
  • README.md
  • hyperalign alignment_paired_t-test.py
  • ridge alphas.py
  • hyperalign; ridge ana.py
  • hyperalign ana_correlation_analysis.py
  • hyperalign anatomical_isc.py
  • hyperalign comp_top_ten_percent.py
  • hyperalign compare_model_fits.py
  • correlation_analysis.py
  • cvu_rdm.pbs : gives location of rdm_isc[life/forward_encoding/rdm_isc.py]
  • docker_r21d_dependencies.sh : installs dependencies for Res(2+1)D - OpenCV, ffmpeg, and Caffe2 based on FB VMZ installation guide in docker/terminal
  • annotate extract_features.py : extracts semantic feature vectors from annotated movie i.e narration/music spectral, narration w2v, image labeling, saliency, motion-energy
  • annotate extract_global_motion.py
  • annotate extract_semantic_category_w2v.py
  • famfaceangles
  • for_sam.py
  • forward.pbs : setting up job for life/forward_encoding/forward_map_corrs.py
  • hyperalign forward_map_corrs.py
  • ridge get_alphas.py : getting alpha value for single run of a participant for leftout hemisphere for one particular model

Heejung's To Do

  • hyperalign get_min_max.py: loads the anatomical-align/ hyper-align dataset and gets the min,max,5%,95% number. Q. what is this number? Hyperaligned beta coefficients?
  • hyperalign get_union_intersection.py: grabs the union of ws and anatomical alignment, grabs the intersection of ws and anatomical alignment. Q. the intersection/union of "vertices", correct?
  • hyperalign group_correlation_test.py: t-test between 1) average fisher z of ws and ha vs 2) average fisherz of ws and aa
  • hyperalign group_smoothness_test.py: using AFNI/SUMA's SurfFWHM, calculates the smoothness in aa, ha, ws. This method is different from the group correlation test and thinking over it would be helpful.
  • one of the files created from this script smoothness_ana.png is currently in /idata/DBIC/cara/models/niml

  • hyperalign hyperalignment.py

  • hyperalign hyperalignment_cvu.pbs: wrapper script for submitting all_ridge.py

  • hyperalign isc isc.pbs: wrapper script for isc.py

  • hyperalign isc isc.py:

  • input:

sam_data_dir = '/idata/DBIC/snastase/life' data_dir = '/idata/DBIC/cara/life/ridge/models' mvpa_dir = '/idata/DBIC/cara/life/pymvpa/' hyperalignmed data: mvpa_dir, 'search_hyper_mappers_life_mask_nofsel_lh_leftout_{0}.hdf5'

  • output:
  1. mv.niml.write(os.path.join(data_dir, '{0}_isc_run{1}_vsmean.lh.niml.dset'.format(model, run)), lh)
  2. isc per runs are stacked and saved as : /idata/DBIC/cara/life/ridge/models/isc mv.niml.write(os.path.join(data_dir, 'isc/{0}_isc_vsmean.lh.niml.dset'.format(model)), np.mean(lh_avg_stack, axis=0)[None,:])
  • annotate **load_files.py **: NOTE - does not seem to be a complete script. load json w2v embeddings for each part (4 files), using json bc has vectors labeled with words
  • input:

json.load(open('/Users/caravanuden/Desktop/life/forward_encoding/old_codes/Part{0}_Raw_Data.json'.format(i)))

  • NOTSURE make_predictions.py
  • input:

does not exist: npy_dir = '/dartfs-hpc/scratch/cara/w2v/w2v_features', but is there something equivalent? motion = np.load('/ihome/cara/global_motion/motion_downsampled_complete.npy')

  • narrative_actions.csv
  • narrative_model.py
  • inputs:

related to this directory? /idata/DBIC/cara/narrative_models
directory = os.path.join('/dartfs-hpc/scratch/cara/new_models/narrative', '{0}/run_{1}'.format(stimfile, fold_shifted), test_p, hemi) -> not quite the same, but /idata/DBIC/cara/new_models/narrative/niml is this similar?

  • output: np.save(os.path.join(directory, 'weights.npy'), wt)
  • annotate narrative_nouns.csv: inputs utilized in what? SHOULD identify the code that calls this csv file in.
  • variance partition_variance.py
  • pca.py: incomplete. reads csv file: l = pd.read_csv(os.path.join('/dartfs-hpc/scratch/cara/w2v/semantic_categories/', f)) Perhaps related to semantic category pca?
  • plot_point_spread.py: incomplete. 1 line of code that loads np.load('/ihome/fma/cara/point_spread_function_results.npz')
  • rdm_isc.py: Probably incomplete? compare with isc.py.
  • input:

sam_data_dir = '/idata/DBIC/snastase/life' suma_dir = '/idata/DBIC/snastase/life/SUMA' mappers = mv.h5load(join(mvpa_dir, 'search_hyper_mappers_life_mask_nofsel_{0}leftout{1}.hdf5'.format(hemi, run))) ds = mv.gifti_dataset(join(sam_data_dir, '{0}_task-life_acq-{1}vol_run-0{2}.{3}.tproject.gii'.format(participant, tr[run], run, hemi)))

  • output:

mv.h5save('/idata/DBIC/cara/search_hyper_mappers_life_mask_nofsel_{0}{1}leftout{1}{2}.hdf5'.format(participant, hemi, left_out, sys.argv[1]), final)

Xiaochun's To Do

  • ridge_regression.py
  • save_for_suma.py
  • save_hyper_data.py
  • save_masked_niml.py
  • save_niml.py
  • save_nuisance.py
  • search_ISC.py
  • search_RDMs.py
  • semantic_categories
  • slh.py
  • slh_combined.lh.niml.dset
  • slh_correlation_analysis.py
  • slh_corrs.lh.niml.dset
  • stats.py: calculates the pair-wise correlation difference of the following models
  • ['ws', 'aa'], ['ws', 'ha_common'], ['aa', 'ha_common'], ['aa', 'ha_testsubj'], ['ws', 'ha_testsubj'], 'ha_testsubj', 'ha_common'
  • ts_corrs.py
  • utils
  • visual_model.py
  • visuals.py
  • annotate w2v.py

2. Semantic Features

Scripts

  • word2vector

Data

  • Behavior
  • Taxonomy
  • Scene

3. Regularized Regression

Scripts

  • Behavior
  • Taxonomy
  • Scene
  • Behavior & Taxonomy
  • Behavior & Scene
  • Taxonomy & Scene
  • Behavior & Taxonomy & Scene

Data

  • Behavior
  • Taxonomy
  • Scene
  • Behavior & Taxonomy
  • Behavior & Scene
  • Taxonomy & Scene
  • Behavior & Taxonomy & Scene

4. Variance Partition

Scripts

  • Behavior
  • Taxonomy
  • Scene
  • Behavior & Taxonomy
  • Behavior & Scene
  • Taxonomy & Scene
  • Behavior & Taxonomy & Scene

Data

  • Behavior
  • Taxonomy
  • Scene
  • Behavior & Taxonomy
  • Behavior & Scene
  • Taxonomy & Scene
  • Behavior & Taxonomy & Scene