TF-MoDISco results

Summary of motifs

TOMTOM matches to motifs

In [1]:
import os
import sys
sys.path.append(os.path.abspath("/users/amtseng/tfmodisco/src/"))
from tfmodisco.run_tfmodisco import import_shap_scores, import_tfmodisco_results
from motif.read_motifs import pfm_info_content, pfm_to_pwm, trim_motif_by_ic
from motif.match_motifs import match_motifs_to_database
from util import figure_to_vdom_image
import plot.viz_sequence as viz_sequence
import numpy as np
import h5py
import matplotlib.pyplot as plt
import vdom.helpers as vdomh
from IPython.display import display

Define constants and paths

In [2]:
# Define parameters/fetch arguments
tf_name = os.environ["TFM_TF_NAME"]
shap_scores_path = os.environ["TFM_SHAP_PATH"]
tfm_results_path = os.environ["TFM_TFM_PATH"]
hyp_score_key = os.environ["TFM_HYP_SCORE_KEY"]
if "TFM_MOTIF_CACHE" in os.environ:
    tfm_motifs_cache_dir = os.environ["TFM_MOTIF_CACHE"]
else:
    tfm_motifs_cache_dir = None

print("TF name: %s" % tf_name)
print("DeepSHAP scores path: %s" % shap_scores_path)
print("TF-MoDISco results path: %s" % tfm_results_path)
print("Importance score key: %s" % hyp_score_key)
print("Saved TF-MoDISco-derived motifs cache: %s" % tfm_motifs_cache_dir)

TF name: MAFK
DeepSHAP scores path: /users/amtseng/tfmodisco/results/importance_scores/multitask_profile/MAFK_multitask_profile_fold2/MAFK_multitask_profile_fold2_imp_scores.h5
TF-MoDISco results path: /users/amtseng/tfmodisco/results/tfmodisco/multitask_profile/MAFK_multitask_profile_fold2/MAFK_multitask_profile_fold2_profile_tfm.h5
Importance score key: profile_hyp_scores
Saved TF-MoDISco-derived motifs cache: /users/amtseng/tfmodisco/results/reports/tfmodisco_results//cache/multitask_profile/MAFK_multitask_profile_fold2/MAFK_multitask_profile_fold2_profile
In [3]:
# Define paths and constants
input_length = 2114
shap_score_center_size = 400
In [4]:
if tfm_motifs_cache_dir:
    os.makedirs(tfm_motifs_cache_dir, exist_ok=True)

Import SHAP scores and TF-MoDISco results

In [5]:
# Import SHAP coordinates and one-hot sequences
hyp_scores, _, one_hot_seqs, shap_coords = import_shap_scores(shap_scores_path, hyp_score_key, center_cut_size=shap_score_center_size)
# This cuts the sequences/scores off just as how TF-MoDISco saw them, but the coordinates are uncut

Importing SHAP scores: 100%|██████████| 311/311 [04:27<00:00,  1.16it/s]
In [6]:
# Import the TF-MoDISco results object
tfm_obj = import_tfmodisco_results(tfm_results_path, hyp_scores, one_hot_seqs, shap_score_center_size)

Plot some SHAP score tracks

Plot the central region of some randomly selected actual importance scores

In [7]:
plot_slice = slice(int(shap_score_center_size / 4), int(3 * shap_score_center_size / 4))
for index in np.random.choice(hyp_scores.shape[0], size=5, replace=False):
    viz_sequence.plot_weights((hyp_scores[index] * one_hot_seqs[index])[plot_slice], subticks_frequency=100)

Plot TF-MoDISco results

Plot all motifs by metacluster

In [8]:
motif_pfms, motif_hcwms, motif_cwms = [], [], []  # Save the trimmed PFMs, hCWMs, and CWMs
motif_pfms_short = []  # PFMs that are even more trimmed (for TOMTOM)
num_seqlets = []  # Number of seqlets for each motif
motif_seqlets = []  # Save seqlets of each motif
metaclusters = tfm_obj.metacluster_idx_to_submetacluster_results
num_metaclusters = len(metaclusters.keys())
if tfm_motifs_cache_dir:
    motif_hdf5 = h5py.File(os.path.join(tfm_motifs_cache_dir, "all_motifs.h5"), "w")
for metacluster_i, metacluster_key in enumerate(metaclusters.keys()):
    metacluster = metaclusters[metacluster_key]
    display(vdomh.h3("Metacluster %d/%d" % (metacluster_i + 1, num_metaclusters)))
    patterns = metacluster.seqlets_to_patterns_result.patterns
    if not patterns:
        break
    motif_pfms.append([])
    motif_hcwms.append([])
    motif_cwms.append([])
    motif_pfms_short.append([])
    num_seqlets.append([])
    motif_seqlets.append([])
    num_patterns = len(patterns)
    for pattern_i, pattern in enumerate(patterns):
        seqlets = pattern.seqlets
        display(vdomh.h4("Pattern %d/%d" % (pattern_i + 1, num_patterns)))
        display(vdomh.p("%d seqlets" % len(seqlets)))
        
        pfm = pattern["sequence"].fwd
        hcwm = pattern["task0_hypothetical_contribs"].fwd
        cwm = pattern["task0_contrib_scores"].fwd
        
        pfm_fig = viz_sequence.plot_weights(pfm, subticks_frequency=10, return_fig=True)
        hcwm_fig = viz_sequence.plot_weights(hcwm, subticks_frequency=10, return_fig=True)
        cwm_fig = viz_sequence.plot_weights(cwm, subticks_frequency=10, return_fig=True)
        pfm_fig.tight_layout()
        hcwm_fig.tight_layout()
        cwm_fig.tight_layout()
        
        motif_table = vdomh.table(
            vdomh.tr(
                vdomh.td("Sequence (PFM)"),
                vdomh.td(figure_to_vdom_image(pfm_fig))
            ),
            vdomh.tr(
                vdomh.td("Hypothetical contributions (hCWM)"),
                vdomh.td(figure_to_vdom_image(hcwm_fig))
            ),
            vdomh.tr(
                vdomh.td("Actual contributions (CWM)"),
                vdomh.td(figure_to_vdom_image(cwm_fig))
            )
        )
        display(motif_table)
        plt.close("all")  # Remove all standing figures
        
        # Trim motif based on information content
        short_trimmed_pfm = trim_motif_by_ic(pfm, pfm)
        motif_pfms_short[-1].append(short_trimmed_pfm)
        
        # Expand trimming to +/- 4bp on either side
        trimmed_pfm = trim_motif_by_ic(pfm, pfm, pad=4)
        trimmed_hcwm = trim_motif_by_ic(pfm, hcwm, pad=4)
        trimmed_cwm = trim_motif_by_ic(pfm, cwm, pad=4)
        
        motif_pfms[-1].append(trimmed_pfm)
        motif_hcwms[-1].append(trimmed_hcwm)
        motif_cwms[-1].append(trimmed_cwm)
        
        num_seqlets[-1].append(len(seqlets))
        
        if tfm_motifs_cache_dir:
            # Save results and figures
            motif_id = "%d_%d" % (metacluster_i, pattern_i)
            pfm_fig.savefig(os.path.join(tfm_motifs_cache_dir, motif_id + "_pfm_full.png"))
            hcwm_fig.savefig(os.path.join(tfm_motifs_cache_dir, motif_id + "_hcwm_full.png"))
            cwm_fig.savefig(os.path.join(tfm_motifs_cache_dir, motif_id + "_cwm_full.png"))
            motif_dset = motif_hdf5.create_group(motif_id)
            motif_dset.create_dataset("pfm_full", data=pfm, compression="gzip")
            motif_dset.create_dataset("hcwm_full", data=hcwm, compression="gzip")
            motif_dset.create_dataset("cwm_full", data=cwm, compression="gzip")
            motif_dset.create_dataset("pfm_trimmed", data=trimmed_pfm, compression="gzip")
            motif_dset.create_dataset("hcwm_trimmed", data=trimmed_hcwm, compression="gzip")
            motif_dset.create_dataset("cwm_trimmed", data=trimmed_cwm, compression="gzip")
            motif_dset.create_dataset("pfm_short_trimmed", data=short_trimmed_pfm, compression="gzip")
if tfm_motifs_cache_dir:
    motif_hdf5.close()

Metacluster 1/2

Pattern 1/10

10843 seqlets

Sequence (PFM)
Hypothetical contributions (hCWM)
Actual contributions (CWM)

Pattern 2/10

822 seqlets

Sequence (PFM)
Hypothetical contributions (hCWM)
Actual contributions (CWM)

Pattern 3/10

274 seqlets

Sequence (PFM)
Hypothetical contributions (hCWM)
Actual contributions (CWM)

Pattern 4/10

231 seqlets

Sequence (PFM)
Hypothetical contributions (hCWM)
Actual contributions (CWM)

Pattern 5/10

199 seqlets

Sequence (PFM)
Hypothetical contributions (hCWM)
Actual contributions (CWM)

Pattern 6/10

91 seqlets

Sequence (PFM)
Hypothetical contributions (hCWM)
Actual contributions (CWM)

Pattern 7/10

86 seqlets

Sequence (PFM)
Hypothetical contributions (hCWM)
Actual contributions (CWM)

Pattern 8/10

86 seqlets

Sequence (PFM)
Hypothetical contributions (hCWM)
Actual contributions (CWM)

Pattern 9/10

48 seqlets

Sequence (PFM)
Hypothetical contributions (hCWM)
Actual contributions (CWM)

Pattern 10/10

47 seqlets

Sequence (PFM)
Hypothetical contributions (hCWM)
Actual contributions (CWM)

Metacluster 2/2

Pattern 1/1

34 seqlets

Sequence (PFM)
Hypothetical contributions (hCWM)
Actual contributions (CWM)

Summary of motifs

Motifs are trimmed based on information content, and presented in descending order by number of supporting seqlets. The motifs are separated by metacluster. The motifs are presented as hCWMs. The forward orientation is defined as the orientation that is richer in purines.

In [9]:
colgroup = vdomh.colgroup(
    vdomh.col(style={"width": "5%"}),
    vdomh.col(style={"width": "5%"}),
    vdomh.col(style={"width": "45%"}),
    vdomh.col(style={"width": "45%"})
)
header = vdomh.thead(
    vdomh.tr(
        vdomh.th("#", style={"text-align": "center"}),
        vdomh.th("Seqlets", style={"text-align": "center"}),
        vdomh.th("Forward", style={"text-align": "center"}),
        vdomh.th("Reverse", style={"text-align": "center"})
    )
)

for i in range(len(motif_hcwms)):
    display(vdomh.h3("Metacluster %d/%d" % (i + 1, num_metaclusters)))
    body = []
    for j in range(len(motif_hcwms[i])):
        motif = motif_hcwms[i][j]
        if np.sum(motif[:, [0, 2]]) > 0.5 * np.sum(motif):
            # Forward is purine-rich, reverse-complement is pyrimidine-rich
            f, rc = motif, np.flip(motif, axis=(0, 1))
        else:
            f, rc = np.flip(motif, axis=(0, 1)), motif
            
        f_fig = viz_sequence.plot_weights(f, figsize=(20, 4), return_fig=True)
        f_fig.tight_layout()
        rc_fig = viz_sequence.plot_weights(rc, figsize=(20, 4), return_fig=True)
        rc_fig.tight_layout()
        
        if tfm_motifs_cache_dir:
            # Save results and figures
            motif_id = "%d_%d" % (i, j)
            f_fig.savefig(os.path.join(tfm_motifs_cache_dir, motif_id + "_hcwm_trimmed_fwd.png"))
            rc_fig.savefig(os.path.join(tfm_motifs_cache_dir, motif_id + "_hcwm_trimmed_rev.png"))

        body.append(
            vdomh.tr(
                vdomh.td(str(j + 1)),
                vdomh.td(str(num_seqlets[i][j])),
                vdomh.td(figure_to_vdom_image(f_fig)),
                vdomh.td(figure_to_vdom_image(rc_fig))
            )
        )
    display(vdomh.table(colgroup, header, vdomh.tbody(*body)))
    plt.close("all")

Metacluster 1/2

#SeqletsForwardReverse
110843
2822
3274
4231
5199
691
786
886
948
1047

Metacluster 2/2

#SeqletsForwardReverse
134

Top TOMTOM matches for each motif

Here, the TF-MoDISco motifs are plotted as hCWMs, but the TOMTOM matches are shown as PWMs.

In [10]:
num_matches_to_keep = 10
num_matches_to_show = 5

header = vdomh.thead(
    vdomh.tr(
        vdomh.th("Motif ID", style={"text-align": "center"}),
        vdomh.th("q-val", style={"text-align": "center"}),
        vdomh.th("PWM", style={"text-align": "center"})
    )
)

for i in range(len(motif_pfms)):
    display(vdomh.h3("Metacluster %d/%d" % (i + 1, num_metaclusters)))
    
    # Compute TOMTOM matches for all motifs in the metacluster at once
    out_dir = os.path.join(tfm_motifs_cache_dir, "tomtom", "metacluster_%d" % i) if tfm_motifs_cache_dir else None
    tomtom_matches = match_motifs_to_database(motif_pfms_short[i], top_k=num_matches_to_keep, temp_dir=out_dir)
    
    for j in range(len(motif_pfms[i])):
        display(vdomh.h4("Motif %d/%d" % (j + 1, len(motif_pfms[i]))))
        viz_sequence.plot_weights(motif_hcwms[i][j])
    
        body = []
        for k, (match_name, match_pfm, match_qval) in enumerate(tomtom_matches[j]):
            fig = viz_sequence.plot_weights(pfm_to_pwm(match_pfm), return_fig=True)
            fig.tight_layout()
            if k < num_matches_to_show:
                body.append(
                    vdomh.tr(
                        vdomh.td(match_name),
                        vdomh.td(str(match_qval)),
                        vdomh.td(figure_to_vdom_image(fig))
                    )
                )
                if tfm_motifs_cache_dir:
                    # Save results and figures
                    motif_id = "%d_%d" % (i, j)
                    fig.savefig(os.path.join(out_dir, motif_id + ("_hit-%d.png" % (k + 1))))
            else:
                body.append(
                    vdomh.tr(
                        vdomh.td(match_name),
                        vdomh.td(str(match_qval)),
                        vdomh.td("Not shown")
                    )
                )
        if not body:
            display(vdomh.p("No TOMTOM matches passing threshold"))
        else:
            display(vdomh.table(header, vdomh.tbody(*body)))
        plt.close("all")

Metacluster 1/2

Motif 1/10

Motif IDq-valPWM
MA1520.1_MAF5.65383e-11
MAFK_HUMAN.H11MO.0.A5.65383e-11
MAFG_HUMAN.H11MO.0.A3.1659100000000005e-10
MA0496.3_MAFK3.4801500000000002e-09
MAFB_HUMAN.H11MO.0.B3.6619400000000003e-09
MA1521.1_MAFA6.16495e-09Not shown
MAF_HUMAN.H11MO.0.A2.6892099999999996e-08Not shown
MAFF_HUMAN.H11MO.0.B6.45307e-08Not shown
MAFK_HUMAN.H11MO.1.A9.70241e-08Not shown
MAFF_HUMAN.H11MO.1.B5.901569999999999e-06Not shown

Motif 2/10

Motif IDq-valPWM
MA0139.1_CTCF8.08285e-17
CTCF_HUMAN.H11MO.0.A1.19475e-12
CTCFL_HUMAN.H11MO.0.A1.93725e-07
MA1102.2_CTCFL0.00017350200000000002
MA1568.1_TCF21(var.2)0.0975277
MA1638.1_HAND20.0975277Not shown
SNAI1_HUMAN.H11MO.0.C0.171552Not shown
ZIC3_HUMAN.H11MO.0.B0.268776Not shown
ZIC2_HUMAN.H11MO.0.D0.450718Not shown
MA1628.1_Zic1::Zic20.451139Not shown

Motif 3/10

Motif IDq-valPWM
MA0117.2_Mafb4.43429e-06
MAFG_HUMAN.H11MO.0.A1.10645e-05
MA0659.2_MAFG1.95382e-05
MA0495.3_MAFF5.6535699999999995e-05
MAFK_HUMAN.H11MO.0.A5.8151000000000006e-05
MAFF_HUMAN.H11MO.0.B0.00338201Not shown
NF2L2_HUMAN.H11MO.0.A0.00904249Not shown
MAF_HUMAN.H11MO.1.B0.00904249Not shown
MA0501.1_MAF::NFE20.00904249Not shown
MA0842.2_NRL0.00904249Not shown

Motif 4/10

Motif IDq-valPWM
MA1102.2_CTCFL0.0217624
WT1_HUMAN.H11MO.0.C0.0217624
MXI1_HUMAN.H11MO.0.A0.0217624
KLF15_HUMAN.H11MO.0.A0.0217624
ZN263_HUMAN.H11MO.0.A0.042020800000000004
KLF16_HUMAN.H11MO.0.D0.0509536Not shown
MA0139.1_CTCF0.0590256Not shown
TBX15_HUMAN.H11MO.0.D0.0590256Not shown
SP3_HUMAN.H11MO.0.B0.0590256Not shown
ZFX_HUMAN.H11MO.1.A0.0590256Not shown

Motif 5/10

Motif IDq-valPWM
MAFK_HUMAN.H11MO.0.A0.00262679
MA0496.3_MAFK0.00262679
MAFG_HUMAN.H11MO.0.A0.00262679
MAF_HUMAN.H11MO.0.A0.00290827
MAF_HUMAN.H11MO.1.B0.00382421
MAFK_HUMAN.H11MO.1.A0.00567786Not shown
MA1520.1_MAF0.0083587Not shown
MA1521.1_MAFA0.0110217Not shown
MAFF_HUMAN.H11MO.0.B0.013826900000000001Not shown
MAFB_HUMAN.H11MO.0.B0.0197955Not shown

Motif 6/10

Motif IDq-valPWM
MA0591.1_Bach1::Mafk1.55284e-05
BACH2_HUMAN.H11MO.0.A2.59181e-05
NFE2_HUMAN.H11MO.0.A6.389630000000001e-05
BACH1_HUMAN.H11MO.0.A9.19017e-05
MA1633.1_BACH19.19017e-05
MA0150.2_Nfe2l20.000272144Not shown
MAFF_HUMAN.H11MO.1.B0.00045814Not shown
MA0501.1_MAF::NFE20.0005555790000000001Not shown
MA0089.2_NFE2L10.000625688Not shown
NF2L2_HUMAN.H11MO.0.A0.000625688Not shown

Motif 7/10

Motif IDq-valPWM
CPEB1_HUMAN.H11MO.0.D4.17826e-06
MA1125.1_ZNF3840.001356
FOXG1_HUMAN.H11MO.0.D0.0175222
FOXL1_HUMAN.H11MO.0.D0.0197173
MA0679.2_ONECUT10.0197173
PRDM6_HUMAN.H11MO.0.C0.0197173Not shown
HXC10_HUMAN.H11MO.0.D0.0197173Not shown
LMX1A_HUMAN.H11MO.0.D0.0412338Not shown
FOXJ3_HUMAN.H11MO.0.A0.0423592Not shown
ARI3A_HUMAN.H11MO.0.D0.0747676Not shown

Motif 8/10

Motif IDq-valPWM
PAX5_HUMAN.H11MO.0.A0.057851400000000004

Motif 9/10

Motif IDq-valPWM
TBX15_HUMAN.H11MO.0.D9.40778e-08
MAZ_HUMAN.H11MO.0.A2.12979e-07
KLF16_HUMAN.H11MO.0.D2.12979e-07
ZN467_HUMAN.H11MO.0.C2.16973e-07
VEZF1_HUMAN.H11MO.0.C1.3968700000000001e-06
SP1_HUMAN.H11MO.0.A1.58953e-06Not shown
PATZ1_HUMAN.H11MO.0.C1.6399599999999997e-06Not shown
SP2_HUMAN.H11MO.0.A2.0714500000000003e-06Not shown
WT1_HUMAN.H11MO.0.C2.7638799999999998e-06Not shown
SP3_HUMAN.H11MO.0.B3.2142700000000003e-06Not shown

Motif 10/10

Motif IDq-valPWM
MA0139.1_CTCF0.00534125
CTCF_HUMAN.H11MO.0.A0.0112055
MA1568.1_TCF21(var.2)0.245575
CTCFL_HUMAN.H11MO.0.A0.364951

Metacluster 2/2

Motif 1/1

No TOMTOM matches passing threshold